Summary

In this chapter, you learned that the models we built in the first few chapters of this book all featured linear growth. You learned that the logistic function was developed to model population growth and then learned how to implement this in Prophet by modeling the growth of the wolf population in Yellowstone after their reintroduction in 1995.

Logistic growth in Prophet can be modeled as either increasing up to a saturation limit called the cap or decreasing to a saturation limit called the floor. Finally, you learned how to model flat (or no growth) trends, where the trend is fixed to one value for the entire data period but seasonality is still allowed to vary. Throughout this chapter, you used the add_changepoints_to_plot function in order to overlay the trend line on your forecast plots.

Choosing the correct growth mode is important, and particularly so when forecasting further into the future. We looked at a couple of examples in this chapter where the incorrect...