Running with Cases: A CBR Approach to Running Your Best Marathon

Abstract

Every year millions of people around the world train for, and compete in, the marathon. As race-day approaches, and training schedulesbegin to wind down, many participants will turn their attention totheir race strategy, as they strive to achieve their best time. To help withthis, in this paper we describe a novel application of case-based reasoningto address the dual task of: (1) predicting a challenging, but achievable,personal best race-time for a marathon runner; and (2) recommendinga race-plan to achieve this time. We describe how suitable cases can begenerated from pairs of race histories and how we can predict a personalbest race-time and produce a tailored race-plan by reusing the race historiesof similar runners. This work is evaluated using data from the lastsix years of the London Marathon.