Modeling Pe2rformance in IRONMAN^® 70.3 Age Group Triathletes.

Area of Science:

• Sports Science
• Data Science
• Endurance Sports Analytics

Background:

• Individual factors influencing age group triathlete performance are documented.
• Limited research exists on applying machine learning (ML) for predicting performance based on these factors.
• This study addresses the gap by using ML to analyze IRONMAN® 70.3 performance predictors.

Purpose of the Study:

• To develop and analyze ML regression models for predicting IRONMAN® 70.3 performance.
• To identify key individual factors (sex, age, country, location) influencing race times.
• To understand complex interactions affecting triathlete performance.

Main Methods:

• Analyzed 823,464 IRONMAN® 70.3 finisher records (2004-2020).
• Utilized triathlete sex, age, country of origin, and event location as predictive factors.
• Built and evaluated four ML regression models, selecting the best performer for further analysis.

Main Results:

• The Random Forest Regressor model demonstrated the highest predictive accuracy.
• Model interpretability revealed that younger men (under 30) from Switzerland or Denmark achieved top performance.
• Optimal race locations included IRONMAN® 70.3 Austria/St. Polten, Switzerland, Sunshine Coast, and Busselton.

Conclusions:

• ML models effectively uncover complex, non-linear relationships influencing triathlete performance.
• These insights can assist IRONMAN® 70.3 age group triathletes in strategic race planning.
• The study highlights the potential of ML in sports performance analysis.