March Madness Metaprediction

In March 2015, I participated in Kaggle’s March Machine Learning Mania—a competition where entrants use machine learning techniques to predict the outcome of NCAA basketball tournament games (I finished a respectable 132nd of 341). As I tracked the results of my predictions through the early rounds, I became interested in the idea of predicting the outcome of the Kaggle competition. This led me to write some R code to run a simulation of the tournament at the beginning of each round to predict the outcome of both the Kaggle competition and the NCAA Tournament.

View the web page I created to display the simulation results here, and the simulation source code here.