Linear Regression on NCAA Basketball Data – Kansas
I went to College Basketball Analysis Powered by SAP Lumira & Predictive Analytics and looked to see if I could do any predictions with the data using Lumira. It turns out because there is no time dimension in the data, I could not. So I went to the Kaggle website that had the data by year.
Kaggle stores the details in one table and the team name in another table, resulting in the merge shown above.
I can then filter and look to see how my team, the University of Kansas, will do.
Last year in 2014 Kansas had a high number of field goals made but didn’t get very far in the tournament. In 2008, Kansas had a low number of field goals yet we won the championship!
The regression projects a downward spiral.
Then looking at three point field goals made, 2014 was high, 2008 was so-so and the linear regression shows in 2015 a projection of 35 3 point field goals, higher than 28 when Kansas won the championship.
Then defense – similar trends with 2008 being low at 58, 2014 at 212 and a projection of 112 rebounds for 2015.
Fouls – a number you want to go down, was high for Kansas in 2014, low in 2008 and is projected to be lower in 2015/2016
Blocked shots – a number you want to go up, was high in 2014 but low in 2008 and projected to be lower in 2015/2016.
I published to Lumira Cloud, but unfortunately the message above is shown. I’ll have to see if there is an idea for this on Idea Place.