My junior year of college, I took Intro To Data Science. At the end of the semester, we had a 3-week long group project. I teamed up with two other Chemical Engineer's and we decided to tackle our favorite sports team, the Utah Jazz. The class had about 80 people in it which made for 20+ groups. At the end of the semester, we submitted our polished projects for a competition to see who was able to take on exciting data and do something productive and clever with it. In a class of lots of different backgrounds, we, the ChemE's, actually won. I know, you don't believe me, but I have proof. Check out http://datasciencecourse.net/2020/fame/ and look at 2018. You'll see my name! Just in case the professors ever take that site down, I'll provide visual evidence.
Okay, well, I guess we were co-winners. But still, we won.
For this project we did the following:
- Scraped an entire season's worth of shot charts from ESPN.com using Beautiful Soup
- Created regular expressions to capture the type of shot from commentary
- Used k-means to create natural clusters of shot locations on the floor.
- Discovered what shooting locations produced the highest expected value for the team as a whole and for individual players
- Performed hypothesis testing to find statistical differences in expected values
- Visualized shooting on the court
- Analyzed differences between home and away games
- Created a predictive model given the shooting locations for a given game.
Here's a fun, 3-minute video explaining what we did and seeing some of our results.
We also of course have our code up on a repo on Github: https://github.com/datascience-course/2018-best-projects/tree/master/Winners/Take%20Your%20Shot%2C%20a%20Shot%20Chart%20Analysis%20of%20the%20Utah%20Jazz
And finally, we have a 91-page report. There are a lot of images and our code is included so really it isn't too long but shows all the detailed results. I can't figure how to add the pdf to the blog yet so I'll have to revisit it.
I'd also like to post at least some of the visuals, even though they aren't the prettiest and I could do much better now days.
All for now.