NBA Win Shares Machine Learning Project

Win Shares is an advanced basketball statistic that allows us to give credit to individual NBA players towards the wins of a team. With a very complex and nuanced formula, Win Shares assigns a fully attributed number to the individual on how much they helped win games. This makes Win Shares a desirable stat for prediction.

In section 1 and 2 of the Analysis Report, I demonstrate how I filtered and adjusted for the data set from the NBA. This particular data set covers 1950-2019, but several filters were applied for this project. On the following page, I reference how the Win Shares formula is calculated and how it is scaled for the NBA. Section 3 is a classification section with how to predict all star selected players from the year 1999.