At the moment, Facebook is a controversial company. However, data scientists can probably agree on this statement: Facebook’s prophet forecasting package is awesome. The main tutorial in the package’s documentation involves forecasting Peyton Manning’s Wikipedia searches. I thought I would do something similar: forecast Mike Trouts’s Wikipedia searches. (Wikipedia data is available from an API for 2015 – present).

Below is a visualization of the forecast for the log of Trout’s Wikipedia searches. The area without the black dots represents the predictions for the future. We see the line captures the general relationship pretty well. However, there are clearly some spikes in interest that would likely not be predictable (e.g. Trout goes 5-5 with a home run).