Associations between internal and external load parameters and match outcomes in men`s volleyball: a machine learning approach

(Zusammenhänge zwischen interner und externer Belastung und Spielausgängen im Männer-Volleyball: ein maschinelles Lernverfahren)

This study aimed to explore the relationship between internal and external loads and their associative value for the success of a professional men`s volleyball team. An observational study involving 11 volleyball athletes from a team in the Portuguese 1st League (age: 20.4 ± 6.34 years). Athletes were monitored throughout the first phase of the 2023/2024 season, encompassing 11 microcycles, 60 training sessions, and 13 matches. An inertial measurement unit was used to measure the number and height of jumps in all data collection contexts, while the rate of perceived exertion (RPE) and session RPE, fatigue, sleep, mood, soreness, stress, and Hooper index was recorded for all sessions and matches during the observation period. The match outcomes (winning or losing) were also recorded for all matches. The logistic regression model achieved an average accuracy score of 93% amongst other metrics, demonstrating a strong ability to capture patterns of winning or losing probabilities. It was found that training duration, RPE, s-RPE, and the number of jumps significantly contributed to the model`s accuracy. Optimizing the workload parameters like training duration, s-RPE, and jump measures may be relevant for improving match outcomes in volleyball, as these factors were closely linked to success in this sample.
© Copyright 2025 International Journal of Performance Analysis in Sport. Taylor & Francis. Alle Rechte vorbehalten.

Bibliographische Detailangaben
Schlagworte:
Notationen:Spielsportarten Naturwissenschaften und Technik
Tagging:maschinelles Lernen Datenanalyse
Veröffentlicht in:International Journal of Performance Analysis in Sport
Sprache:Englisch
Veröffentlicht: 2025
Online-Zugang:https://doi.org/10.1080/24748668.2024.2442862
Jahrgang:25
Heft:4
Seiten:707-722
Dokumentenarten:Artikel
Level:hoch