Runner's Performance Predictor

As an ex-competitive runner, I can appreciate an application that predicts my expected future performance based on my recent training and racing history.

This can be accomplished with machine learning techniques and your individual historical performance. This application is a proof of concept for exactly that customized type of service.

Using competitors' historical data obtained from the USATF site, we created a predictive model that provides a goal performance to make it to the All-time Masters runner list.

Competitors' predictors, which include: Previous Race Performance, Age, and Gender provided the basis for this application predictive services.

While this application generalizes across the field of competitors, there is no reason why a similar methodology would not be possible to provide this service for a particular individual.

The following application is a small example of the services we can provide for your company

load("cubist_mod.Rda") # our saved cubist model

# predict minutes a 65 year-old male runner in the USATF All-time List will perform in a 10K race
predict(cubist.mod, data.frame(Male=1 ,  Meters=10000 , Age=65))/60

[1] 33.87344