Beginner

Intermediate

Advanced

Explore comprehensive Machine Learning interview questions and answers with detailed explanations.

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Bias<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">: Error caused by overly simple assumptions in a machine learning model. It usually comes from models that are too rigid.<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Variance<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">: Error caused by sensitivity to small changes in training data. It increases when models learn noise instead of patterns.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">High Bias<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> leads to <span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">underfitting<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> and missed patterns.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">High Variance<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> leads to <span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">overfitting<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> and poor generalization.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Explain the bias-variance tradeoff. How does it relate to model performance?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Supervised Learning: This approach learns from input–output pairs.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Uses <span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">labeled datasets<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> with known outputs.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Common tasks: Classification and regression.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Examples: Email spam class ification, house price prediction.<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Unsupervised Learning: This approach works without target labels.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Uses <span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">unlabeled datasets<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Finds hidden structures or patterns.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Examples: Customer segmentation, anomaly detection.<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is the difference between supervised and unsupervised learning? Provide examples of each.

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Definition<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">: Overfitting occurs when a model memorizes training data instead of learning general patterns. <span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">It reduces real-world performance.<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Signs of Overfitting<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Very high training accuracy.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Low validation or test performance.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Prevention Techniques<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Regularization methods.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Cross-validation.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Early stopping during training.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Reducing model complexity.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Increasing dataset size.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is overfitting? How can you detect and prevent it?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Underfitting: It happens when the model is too weak to learn.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Model is too simple.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Performs poorly on both training and testing data.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Indicates high bias.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Overfitting: It occurs when the model becomes overly detailed.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Model is too complex.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Performs well on training data but poorly on new data.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Indicates high variance.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Explain what underfitting means and how it differs from overfitting.

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Parameters<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Learned automatically during training.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Examples: Model weights, coefficients.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Hyperparameters<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Set before training begins.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Control the learning process.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Examples: Learning rate, number of epochs, tree depth.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is the difference between a parameter and a hyperparameter?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Purpose<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Train the model on one portion of data.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Evaluate performance on unseen data.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Benefits<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Measures generalization ability.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Prevents biased evaluation.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Typical Ratios<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">80% training / 20% testing.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">70% training / 30% testing.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is the purpose of train-test split? What is a typical split ratio?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Validation Set<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -13.5pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Used for tuning hyperparameters and selecting models.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Test Set<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -13.5pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Used only for final unbiased evaluation.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Why do we need a separate Validation set if we already have a Test set?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Cross-validation<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> evaluates a model using multiple training and validation splits. It reduces dependence on one random split.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Helps reduce randomness from a single split.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Produces more reliable and stable performance estimates.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is cross-validation and why is it important?

Dataset divided into <span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">k equal parts (folds). Each fold acts as validation once.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Model trained <span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">k times<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">, each time using a different fold for validation.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Remaining folds used for training.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Final result calculated by averaging all evaluation scores.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Explain what k-fold cross-validation is and how it works.

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Parametric Models<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Fixed number of parameters.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Faster training and lower complexity.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Example: Linear regression.<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Non-Parametric Models<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Flexible structure without fixed assumptions.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Can model complex relationships.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Examples: Decision trees, k-nearest neighbors. <span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is the difference between parametric and non-parametric models?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Definition:<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> Linear regression predicts a continuous value by fitting a straight line that represents the relationship between input variables and the target. It is one of the simplest supervised learning algorithms.<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">How it Works<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Finds the best-fit line by minimizing the difference between predicted and actual values.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Uses coefficients (weights) and an intercept.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -9pt;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Commonly optimized using least squares error.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Key Assumptions<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Linear relationship between features and target.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Errors are independent.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Constant variance of errors (homoscedasticity).</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Errors are normally distributed.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Low multicollinearity between features.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Explain how a linear regression model works. What assumptions does it make?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">L1 Regularization (Lasso)<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Adds absolute value of coefficients as a penalty.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Can shrink some coefficients exactly to zero.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Useful for feature selection.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Produces sparse models.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">L2 Regularization (Ridge)<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Adds squared value of coefficients as a penalty.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Reduces coefficient size but rarely makes them zero.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Helps reduce overfitting.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Keeps all features in the model.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is the difference between L1 (Lasso) and L2 (Ridge) regularization?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Linear Regression<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Used for predicting continuous values.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Output can be any real number.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Uses a straight line relationship.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Logistic Regression<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Used for classification problems.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Output is a probability between 0 and 1.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Applies a sigmoid function to produce class probabilities.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

How does logistic regression differ from linear regression?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Reason:<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> Logistic regression predicts probabilities using a regression-style mathematical model. It estimates continuous values before final classification.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;It models the relationship between inputs and the log-odds of the outcome.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Even though the final task is classification, the underlying method estimates continuous probability values.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">The name comes from estimating a regression equation before applying a classification threshold.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Why is logistic regression called "regression" when it's used for classification?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Purpose:<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> The sigmoid function converts any real-valued input into a probability between 0 and 1. It makes the output suitable for binary classification.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Maps linear output into a bounded range.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Allows interpretation as probability.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Helps create a clear decision boundary for classification.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Makes the model suitable for binary outcomes.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is the purpose of a sigmoid function in logistic regression?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Classification<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Predicts discrete class labels.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Output is category like yes or no.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Used for spam detection or disease prediction.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Regression<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Predicts continuous numeric values.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Output can be any real number.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Used for price or temperature prediction.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is the difference between classification and regression?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Key Assumptions<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Linear relationship between input and output variables.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Errors are independent from each other.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Constant variance of residuals (homoscedasticity).</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Errors follow normal distribution.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Features have low multicollinearity.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Explain the assumptions of Linear Regression.

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Ridge Regression<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Used when all features are important.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Works well when features are highly correlated.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Reduces coefficient size without removing features.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Lasso Regression<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Used when feature selection is needed.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Can make some coefficients zero.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Produces simpler sparse models.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

When would you use Ridge regression over Lasso regression?

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Confusion Matrix:<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> A table used to evaluate classification model performance. It compares predicted labels with actual labels.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;True Positive (TP) correct positive predictions.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">True Negative (TN) correct negative predictions.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">False Positive (FP) wrong positive predictions.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">False Negative (FN) wrong negative predictions.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Explain the concept of a confusion matrix and its components.

<span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Precision<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Measures how many predicted positives are correct.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Focuses on prediction quality.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Important when false positives are costly.</li></ul><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Recall<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Measures how many actual positives are found.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Focuses on coverage of positives.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:'Times New Roman',serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Important when missing positives is risky.</li></ul>

What is the difference between precision and recall?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Understand the Data First<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Remove irrelevant or duplicate features.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Handle missing values and scaling.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Use Statistical Methods<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Correlation analysis.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Mutual information or chi-square tests.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Apply Model-Based Techniques<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Regularization models such as Lasso.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Tree-based feature importance.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Dimensionality Reduction<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">PCA or feature embedding techniques.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

How would you approach feature selection for a high-dimensional dataset?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Filter Methods<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Select features using statistical tests.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Fast and model-independent.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Example: Correlation threshold.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Wrapper Methods<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Use model performance to evaluate feature subsets.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">More accurate but computationally expensive.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Example: Recursive Feature Elimination.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Embedded Methods<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Feature selection happens during model training.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: -4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Balanced between speed and accuracy.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Example: L1 regularization.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is the difference between filter, wrapper, and embedded methods for feature selection?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Feature Importance<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Measures how much each feature contributes to predictions.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">In Tree-Based Models<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Calculated from reduction in impurity (Gini or entropy).</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Features used for better splits get higher scores.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Benefit<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Helps interpret models and remove less useful features.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Explain the concept of feature importance. How is it calculated in tree-based models?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Definition<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 13.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Measures importance by shuffling feature values.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Process<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 13.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Randomly permute one feature at a time.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 13.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Observe drop in model performance.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Insight<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 13.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Large performance drop means the feature is important.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 13.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Model-agnostic and easy to interpret.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is permutation importance and how does it work?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Avoid Basic One-Hot Encoding<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Prevents very large feature spaces.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Use Advanced Encoding<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Target encoding.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Frequency encoding.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Embedding techniques.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Feature Engineering<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Group rare categories into an “other” class.</li></ul> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

How would you handle categorical features with high cardinality?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Target Encoding<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Replaces categories with the mean of the target variable.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Advantages<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Handles high-cardinality features efficiently.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Reduces dimensionality.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Risks<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Data leakage.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Overfitting if applied without cross-validation.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is target encoding, and what are its risks?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Detect Multicollinearity<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Correlation matrix.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Variance Inflation Factor (VIF).</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Solutions<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Remove highly correlated features.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Use dimensionality reduction like PCA.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Apply regularization techniques.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Impact<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Improves model stability and interpretability.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

How would you handle multicollinearity in your features?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Definition<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Measures how much a feature is correlated with other features.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Interpretation<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">VIF = 1: No correlation.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">VIF &gt; 5 or 10: High multicollinearity risk.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Usage<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Calculate VIF for each feature.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;margin-left: 4.5pt;" aria-level="2"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Remove or combine features with high VIF values.</li></ul> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is the VIF (Variance Inflation Factor) and how do you use it?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Data Leakage<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> happens when model learns from <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">future/test data<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">. It causes unreal performance during validation. It usually occurs during <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">feature engineering<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">, <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">scaling<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">, or wrong <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">data splitting<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Use <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">train-test split<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> before preprocessing.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Apply <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">pipelines<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> for transformations.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Remove <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">target leakage<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> features.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Use correct <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">cross-validation<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Explain the concept of data leakage. How can you prevent it?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Imbalanced data biases model toward <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">majority class<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">. Accuracy alone becomes misleading. Using metrics like <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">F1-score<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> helps measure minority performance.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Apply <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">SMOTE/ADASYN<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Use <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">class weights<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Evaluate with <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">F1 / ROC-AUC<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Try <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">ensemble methods<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

How would you handle a highly imbalanced dataset in classification?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">SMOTE<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> and <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">ADASYN<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> generate synthetic minority samples. They help when class distribution is highly uneven. They reduce bias toward <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">majority class<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> during training.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">SMOTE<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> uses nearest neighbors.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">ADASYN<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> focuses on hard samples.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Works for <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">imbalanced classification<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Avoid with heavy <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">noise<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What are SMOTE and ADASYN? When would you use them?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">MNAR<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> means missingness depends on hidden causes. Simple imputation can introduce bias. Adding indicators helps model learn missing patterns.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Add <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">missing indicator<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> feature.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Use <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">model-based imputation<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Perform <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">data analysis<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> first.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Try <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">multiple imputation<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

How would you handle missing data that is not random (MNAR)?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Stratified Sampling<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> keeps same class ratio in splits. It ensures fair evaluation for imbalanced datasets. It reduces randomness that may distort evaluation results.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Maintain <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">class distribution<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Reduce <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">sampling bias<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Use in <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">StratifiedKFold<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Important for <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">imbalanced data<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is stratified sampling and when it's necessary?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Outliers are extreme values affecting model learning. They may come from errors or rare events. Not all outliers should be removed; some contain useful signals.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Use <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">IQR<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> or <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Z-score<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Visualize with <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">boxplot<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Apply <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">log transform<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Remove only if <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">invalid<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

How would you detect and handle outliers in your dataset?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Noisy Labels<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> reduce model performance. They occur due to human or system errors. Cleaning labels improves generalization and stability.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Use <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">robust loss<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> functions.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Apply <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">data cleaning<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Train with <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">ensembles<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Filter by <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">confidence score<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

How do you handle noisy labels in supervised learning?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Online Learning<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> updates model continuously with new data.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Batch Learning<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> trains using full dataset at once.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Online Learning vs Batch Learning<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> — <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Online<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> works for streaming data and fast updates. <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Batch<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> is stable but needs full retraining.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Online supports <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">streaming data<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Batch gives <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">stable training<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Online uses <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">incremental updates<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Batch needs full <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">retraining<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

What is the difference between online learning and batch learning?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Both are <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">ensemble tree models<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> but learn differently.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Random Forest<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> builds trees independently.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Gradient Boosting<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> learns sequentially to reduce errors.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Random Forest vs Gradient Boosting<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> — <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Random Forest<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> is faster and less sensitive to tuning. <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Gradient Boosting<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> gives higher accuracy with careful tuning.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Random Forest<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> reduces variance.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Gradient Boosting<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> reduces bias.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Choose RF for <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">fast baseline<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Choose GB for <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">high accuracy<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Compare Random Forest and Gradient Boosting. When would you choose one over the other?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">XGBoost<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> is optimized and regularized boosting algorithm. It improves speed and prevents overfitting.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Traditional Gradient Boosting vs XGBoost<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> — <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">XGBoost<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> adds regularization, parallelism, and better tree pruning.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Includes <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">regularization<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Supports <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">parallel processing<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;Handles <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">missing values<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Uses <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">tree pruning<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

How does XGBoost improve upon traditional gradient boosting?

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Regularization controls model complexity in boosting. It reduces overfitting and improves generalization.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Alpha vs Lambda<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> — <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Alpha (L1)<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> encourages sparsity while <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Lambda (L2)<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> controls large weights.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">&nbsp;<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Alpha (L1)<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> adds sparsity.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Lambda (L2)<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> controls weights.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Improves <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">generalization<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Stabilizes training.</li></ul><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">

Explain the concept of regularization in XGBoost (alpha and lambda parameters).

<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Model choice depends on data patterns and goals. <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Linear Models<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> assume simple relationships. <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Tree-Based Models<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> capture complex interactions.<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Linear Models vs Tree-Based Models<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> — <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Linear<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> models are interpretable and fast. <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Tree Models<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> handle non-linear data and feature interactions better.<ul style="margin-top:0;margin-bottom:0;padding-inline-start:48px;"><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Use <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">linear models<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;"> for simple trends.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Use trees for <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">non-linear data<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Linear gives better <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">interpretability<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li><li dir="ltr" style="list-style-type:disc;font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;" aria-level="1"><span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">Trees handle <span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:700;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">feature interactions<span style="font-size:11pt;font-family:Arial,sans-serif;color:#000000;background-color:transparent;font-weight:400;font-style:normal;font-variant:normal;text-decoration:none;vertical-align:baseline;white-space:pre;white-space:pre-wrap;">.</li></ul>

How do you choose between a linear model and a tree-based model?