My name is Elan Rotenberg. I grew up in Baltimore, Maryland, but now live in New York City. I recently graduated from Yeshiva University with a Masters in applied mathematics. I had been working part time as a math teacher and math tutor. I completed two theses while at Yeshiva University, coding them both in Mathematica, giving me a passion for it. One was about powerlifting and the other about basketball (I love sports). Outside of school and work, I spend much of my time playing or watching sports with friends. I am very excited to be a part of the 2019 Wolfram Summer School. Looking forward to meeting everyone!
Computational Essay: How Strong are Giants?
The ability to be able to predict games with good accuracy delivers many benefits. Using data from the last 10 years on teams’ scoring, rebounds, assists, free throws and Elo ratings (calculation of the relative skill level of a team), I was able to build a machine learning algorithm that accurately predicted 65.4% of the 2018–2019 NBA season. This performance is closely aligned with that of top data analytics companies.
Summary of Results
I imported data from NBA.com and structured it in an organized fashion. I then performed calculations on this data. I then predicted the 2018–2019 NBA season by using this data of successive NBA seasons as a training set. After four seasons, the training data no longer improved performance. Several combinations of the data were used to predict the outcome, the highest one being Elo and location. Using only these two variables, I was able to build a machine learning algorithm that performs as well as that of top data analytics companies.
1. Consider external factors such as injured players, rested players, travel time, back-to-back games, etc. 2. Make a cloud version that updates automatically. 3. Take expected lineups into account.