Holds all of the files related to Anthony and Maxwell Holloway's fantasy baseball prediction project.
The basic idea of this project is simple. My father and I thought that there would be potential for machine learning to predict the performance of a player on a day-to-day basis. Standard guessing methods may estimate a player's worth by taking his average performance over the season, however this is missing all of the nuance of their performance based on recorded factors, such as the pitcher, the time of year, the weather, the location (home vs away), and others.
In an effort to make educated guesses on the performance of the players, we developed a simple model.
1.) Train a neural net for every batter, using a training set taken from every game they've played (using both the Lahman dataset and retrosheet.org). 2.) Use a solving tool in excel to find which combination of players will maximize the number of points scored while remaining under the salary cap of the service.
Although this is a simple process, it will hopefully provide at least marginally better results than computing the average.