A machine learning approach to finding the fastest race course for professional athletes competing in Ironman® 70.3 races between 2004 and 2020

• Published in: International journal of environmental research and public health 2023-02, Vol.20 (4), p.1-12
• Main Authors: Thuany, Mabliny, Valero, David, Villiger, Elias, Miguel Forte, Pedro, Weiss, Katja, Nikolaidis, Pantelis Theo, Andrade, Marília S., Cuk, Ivan, Sousa, Caio Victor, Knechtle, Beat
• Format: Article
• Language: English
• Subjects: Algorithms; Athletes; Athletic Performance; Bicycling; Cycling; Decision trees; Endurance; Gender; Humans; Independent variables; Learning algorithms; Machine Learning; Male; Physical Endurance; Races; Regression analysis; Running; Swimming; Triathlon
• ISSN: 1660-4601; 1661-7827; 1660-4601
• DOI: 10.3390/ijerph20043619

Our purpose was to find the fastest race courses for elite Ironman® 70.3 athletes, using machine learning (ML) algorithms. We collected the data of all professional triathletes competing between 2004 and 2020 in Ironman 70.3 races held worldwide. A sample of 16,611 professional athletes originating from 97 different countries and competing in 163 different races was thus obtained. Four different ML regression models were built, with gender, country of origin, and event location considered as independent variables to predict the final race time. For all the models, gender was the most important variable in predicting finish times. Attending to the single decision tree model, the fastest race times in the Ironman® 70.3 World Championship of around ~4 h 03 min would be achieved by men from Austria, Australia, Belgium, Brazil, Switzerland, Germany, France, the United Kingdom, South Africa, Canada, and New Zealand. Considering the World Championship is the target event for most professional athletes, it is expected that training is planned so that they attain their best performance in this event.