Using Neural Networks to Predict Cross Country and Track Race Times

Project Description

Cross country and track races are governed by many factors, and it is impossible to predict exactly what will happen on any given day. On one weekend, one person will win, and the next weekend, with almost the exact same competitiors and course, someone else will come out victorious. Despite the difficulties posed in predicting races, many runners spend hours pouring over race results and prediction websites. One recently developed prediction site is www.lacctic.com, which uses a weighting algorithm to compare runners based on previous performances.

This project will train neural networks to predict the results of a race using previous race results, course information, weather conditions, time of day, etc. Race data is freely accessible online at www.athletic.net and www.tfrrs.org. Additional data will need to be collected about general race and course inforrmation.

As a runner and someone who obsesses over race results, I hope that this project could help people to see how they compare, set achievable goals, metally prepare for race day, and see if there are any hidden factors influencing race results.

Project Goals

• Train a neural network to predict the results of a cross country or track race of a specific distance
• Expand the NN to perdict the results of races of any distance
• Determine if there are any hidden factors that influence race results
• Explore the ethical implications of these sorts of predictions, and how this could change the sport