Building A Stanley Cup Winning Roster Using a K-Nearest Neighbors Algorithm
The goal of every NHL general manager is to construct a Stanley Cup winning roster. However, due to the limited availability of analytics tools, this task is extremely complex. In an attempt to simplify this process, a k-nearest neighbors algorithm was created, classifying players as one of four line-type players, using publicly available player evaluation statistics. The algorithm was modified to classify players as first liners, second liners, or bottom sixers, resulting in increased accuracy. Finally, the algorithm was modified to classify players as either top or bottom six players, leading to much higher accuracy. With further development, this algorithm opens the doors for increased confidence in player trades, drafting, and contract negotiations.
Visuals
Created for DTSA 5304: Fundamentals of Data Visualization, this project was an assignment meant to learn how to visualize the results of an analytics project.
Visuals is currently under construction. See my blog to learn about how I'm updating this project.