Every collegiate basketball team in the country wants a chance to play in the NCAA Men’s Basketball tournament. Only 68 teams make it, hence the excitement and thrill around this event. For seventy-seven years, the highly anticipated NCAA Division I collegiate basketball tournament has taken place. From the University of Oregon winning in 1939, to Villanova claiming the title in last year’s competition, there has been plenty of excitement throughout the years surrounding this event. Given the event’s highly anticipated popularity and excitement, a prescriptive model is being created in an attempt to accurately predict the outcome of the upcoming tournament. To help with this matter, a preliminary analysis will be performed to determine the model’s accuracy, and it will be monitored as the tournament progresses. Further analysis will include incorporating advanced statistics such as:
- SOS Difference, which stands for strength of schedule measuring the quality of competition for each team.
- Location, whether the winning team was playing at home, away or on a neutral court.
- Three Point Reliance, which explains the percentage of points scored from three pointers for a team. Part of the objective for implementing this project is to apply most of the skills acquired throughout the program in a setting that is applicable to both college and non-college students, as well as sports and non-sports fans. By creating this model, one can understand the complexities that exist in such an assignment, and one can further understand all the work that is required for other related assignments that exist across different industries. Plus, given that this galvanizes people and there is a lot of stake, individuals who attempt to create such model can get closer to a potential breakthrough in the world of analytics that may potentially shed some light into other analytics-related issues that are present in other sectors.
To see this notebook live, please head on over to: https://nccab.notebook.us-east-1.sagemaker.aws/notebooks/WP_Blog_CBB.ipynb
The link above has the code for the model as well as any pertinent comments for each code block.
If you would like to demo our model then please proceed to: https://criviere.github.io/ncaa_camilo_armando
At the demo link you will be able to select matchups for any college team of your preference and have a prediction returned on the teams of your choice.