Weather-aware prediction of trail running finish times using machine learning

Darunnaja, Muhammad Annas, Hariyadi, M. Amin ORCID: https://orcid.org/0000-0001-9327-7604, Aziz, Okta Qomaruddin, Prakasa, Johan Ericka Wahyu ORCID: https://orcid.org/0000-0001-5571-9328 and Almais, Agung Teguh Wibowo (2026) Weather-aware prediction of trail running finish times using machine learning. International Journal of Advances in Data and Information Systems, 7 (2). pp. 738-746. ISSN 2721-3056

Text 29002.pdf - Published Version Available under License Creative Commons Attribution Share Alike. Download (804kB)

Abstract

This study investigates the role of environmental variables in improving the prediction of Mean Finish Time (MFT) in trail running events. While previous approaches primarily rely on track-related features to predict individual athlete performance, the contribution of dynamic weather conditions at the event level remains insufficiently explored. This research adopts a quantitative modeling approach using Extreme Gradient Boosting (XGBoost) to analyze 36,700 race records integrated with spatio-temporal weather data. To rigorously prevent data leakage, a controlled experimental design was implemented using Group Shuffle Split based on race titles, comparing a model that incorporates environmental variables against one relying solely on track characteristics. The results show that the inclusion of weather variables significantly improves predictive reliability, reducing the Mean Absolute Percentage Error (MAPE) from 10.05% to 8.50% and increasing the coefficient of determination ( ) to 0.7836. Further analysis reveals that environmental variables, particularly temperature, interact with terrain difficulty and disproportionately influence high-effort events. In conclusion, integrating environmental variables significantly enhances predictive accuracy, offering a novel, data-driven approach for race organizers to calculate ideal Cut-Off Times (COT) and Cut-Off Points (COP) based on dynamic environmental constraints.

Item Type:	Journal Article
Keywords:	trail running; extreme gradient boosting; explainable ai; event-based modeling; weather impact
Subjects:	08 INFORMATION AND COMPUTING SCIENCES > 0801 Artificial Intelligence and Image Processing > 080105 Expert Systems
Divisions:	Faculty of Technology > Department of Informatics Engineering
Depositing User:	Mokhamad Amin Hariyadi
Date Deposited:	24 Sep 2026 09:15

Downloads

Origin of downloads

Actions (login required)

View Item