Including Additional Regressors
In your first model in Chapter 2, Getting Started with Prophet, you forecasted carbon dioxide levels at Mauna Loa using only the date (but no other information) to predict future values. Later, in Chapter 6, Forecasting Holiday Effects, you learned how to add holidays as additional information to further refine your predictions of bicycle ridership in the Divvy bike share network in Chicago.
The way holidays are implemented in Prophet is actually a special case of adding a binary regressor. In fact, Prophet includes a generalized method for adding any additional regressor, both binary and continuous.
In this chapter, you’ll enrich your Divvy dataset with weather information by including it as an additional regressor. First, you will add binary weather conditions to describe the presence or absence of sun, clouds, or rain, and then you will bring in continuous temperature measurements. Using additional regressors can allow you to include...
