The association of origin and environmental conditions with performance in professional IRONMAN triathletes.

We have (i) little knowledge about where the fastest professional IRONMAN triathletes originate from and where the fastest races take place and (ii) we have no knowledge of the optimal weather conditions for an IRONMAN triathlon. The aims of the present study were, therefore, (i) to investigate the origin and the fastest IRONMAN race courses for professional triathletes and (ii) to evaluate the best environmental conditions (i.e. water and air temperatures and type of race course) for the fastest IRONMAN race times in professional IRONMAN triathletes. Data of all professional female and male IRONMAN triathletes competing between 2002 and 2022 in all IRONMAN races held worldwide were collected. A total of 6,943 finishers´ records (4,162 from men and 2,781 from women) from 58 different countries participating in 54 different event locations between 2002 and 2022 were considered. Data was analyzed using descriptive statistics and machine learning (ML) regression models. The models considered gender, country of origin, event location, water, and air temperature as independent variables to predict the final race time. Three different ML models were built and evaluated, based on three algorithms, in order of growing complexity and predictive power: Decision Tree Regressor, Random Forest Regressor, and XG Boost Regressor. Most of the athletes originated from the USA (1786), followed by athletes from Germany (674), Canada (426), Australia (396), United Kingdom (342), France (325), and Switzerland (276). Most of the athletes competed in IRONMAN Hawaii (925), IRONMAN Florida (563), IRONMAN Austria (452), IRONMAN France (354), IRONMAN Wisconsin (330), IRONMAN Lanzarote (322) and IRONMAN Texas (313). The Decision Tree and the XG Boost models were the best performing models (r