Using ternary search to find the best fitting model and minimize the loss function for NFL winrate machine learning.
This project import the data from the entire 2016 NFL season along with the first six weeks of the 2017 NFL season. This project will use the 2016 data to determine the penalty coefficient for the model that makes the best predictions. We then fit the model with that penalty coefficient using the 2017 data in order to make predictions for weeks seven and later of the 2017 season.