Appendix B: Forecasting Methods

This appendix describes the supported forecasting methods.

Simple Exponential Smoothing

This forecasting method develops a model where the prediction is a weighted linear sum of recent past observations. Exponential Smoothing models explicitly use an exponentially decreasing weight when considering past observations. Single Exponential Smoothing only requires one parameter alpha, also called the smoothing factor or smoothing coefficient. Alpha controls the rate of influence of the observations at each prior timestep. The parameter is handled inside of the algorithm and thus is unspecified by a user.

NOTE: Simple Exponential Smoothing Models are best at handling data with no clear trend or seasonal components.