malcolmgaynor/a-case-study-in-olympic-weightlifting

Using logistic regression, decision trees, and Markov chains

A-case-study-in-Olympic-weightlifting

Project with Liam Rosengren, as a part of the Stat 306: Multivariate Sports Analytics class with Professor Bradley A. Hartlaub at Kenyon College. April 29th, 2024.

In this project, Liam and I analyzed Olympic weightlifting strategy, specifically the strategy of weightlifter Caine Wilkes, who competed for the USA in the 2020 Tokyo Olympic Games in the heaviest category, 109kg+. First, we used logistic regression models to show the importance of body weight and age in lift success rate. Also, we looked into the correlation between weight and age and the performance of all ultra heavyweight (109kg+) weightlifters, demonstrating the importance of bodyweight for these athletes.

Next, we created a decision tree to help predict whether or not Caine Wilkes would successfully complete a lift, depending on factors such as weight attempted, bodyweight, age, etc. Then, we used decision trees to create multiple Markov chains, each representing potential strategies Wilkes could use if he competes in the 2024 Paris Olympic Games. These Markov chains were computed to find the probability that Wilkes finishes with various different amounts of weight lifted, depending on the strategy he chooses. Details about the strategies we considered and the computations can be found in our presentation slides.

This repository includes our Google Slides presentation sharing the details behind our process, our two page executive summary of results, the code (written in R) used to create our models and do the analysis, and the data we considered. If you have any questions or are interested in our process, data, models, code, or analysis, please do not hesitate to reach out!