This notebook contains 6 different type of models with roughly ~10 different type of model fitting algorithms fit for best accuracy. There is 3 models to predict YoY of WAR (Wins above Replacement), GAR (Goals above replacement), and SPAR (Standing Points above replacement) for Forwards and 3 of the same type of models for defensemen. This was used for Northwestern University Capstone Project for Masters of Science in Data Science
The goal is to help identify players that are likely to have an improvement WAR, GAR, or SPAR YoY. The results of the model ended up being able to classify those players with high probability of a bounce back year.
The results of the model are included in the following publication: https://arch.library.northwestern.edu/concern/parent/ks65hc477/file_sets/qf85nb595
The application of the model allows Hockey Operations to identify players who may have been cheap relative to their recent season's performance but are likely to have an improved following season.
Data downloaded from EvolvingHockey.com
Link to dashboard: https://public.tableau.com/profile/sean.prentis#!/vizhome/MSDS_NHL_Capstone/TitlePage
