Factor and Correlation Analysis for Predicting Marathon Race Performance Using Machine Learning Algorithms

Abstract

A marathon race is a long-distance running event typically spanning 26.2 miles or 42.195 kilometers. It is a test of endurance, stamina, and mental fortitude, attracting participants from all walks of life, ranging from elite athletes to recreational runners. The origins of the marathon can be traced back to ancient Greece, where legend has it that a messenger named Pheidippides ran from the battlefield of Marathon to Athens to deliver news of victory before collapsing and dying from exhaustion. Machine learning has increasingly become a valuable tool in optimizing training strategies, performance prediction, and injury prevention for marathon runners. By analyzing vast amounts of data collected from wearable devices, training logs, and race results, machine learning algorithms can identify patterns, trends, and correlations that help runners improve their training regimens and race-day strategies. This paper introduces a novel approach, the Factor Analysis Probabilities Prediction Ranking Machine Learning (FA-PP-R-ML) methodology, for predicting marathon race outcomes. With historical race data, factor analysis, and machine learning techniques, the FA-PP-R-ML methodology aims to accurately estimate marathon finish times and rank predicted outcomes based on their probabilities. Through a comprehensive analysis of marathon race data, including training metrics, environmental conditions, and physiological parameters, the FA-PP-R-ML model identifies latent factors influencing race performance. Through factor analysis, latent factors influencing race performance are identified, with values ranging from 0.5 to 0.9. Machine learning algorithms utilize these factors to predict marathon finish times, resulting in accurate predictions with an average error of ±0.1 hours.