Using Machine Learning to Build a Better Fitness App to Help Runners to Run a Faster Marathon

We explore using machine learning to help marathoners achieve a personal best for an upcoming race, by helping them to select a goal-time and a pacing plan. We evaluate several representational alternatives, and algorithms, using real-world race data, to highlight the performance implications of different types of marathon histories and landmark races, concluding that richer representations do not always deliver better prediction performance.