Time Series Analysis with Facebook Prophet Python and Cesium

This project leverages the Prophet model with Cesium-extracted features to enhance healthcare call volume forecasting accuracy. The reusable forecasting pipeline can be applied to various industries for better demand prediction and decision-making.

• Using the Prophet model, accurately forecasted healthcare call volumes for the next 12 months.

• Improved model accuracy by adding additional statistical features such as mean, standard deviation, and absolute difference.

• To enhance the predictions, we add Cesium-extracted features as additional regressors to the Prophet model.

• Better forecast trends, seasonality, and uncertainty intervals are visualized for better decision-making.

• They were able to identify key patterns and trends in healthcare call data - such as seasonal peaks and troughs.

• Forwards fill without breaking dataset continuity, to help handle missing data.

• They generate reliable predictions and can be used to guide resource allocation and staffing needs of healthcare facilities.

• Helped us understand what drove call volumes-seasonality, outliers, and all.

• Built a reusable forecasting pipeline that can be used in other sectors or on other datasets.

• It also demonstrated how combining Prophet and Cesium provides better time series forecasting accuracy.