Main variables predicting motor skills: analysis with classification trees; [Principales variables que predicen la competencia motora: análisis con árboles de clasificación]

Abstract

Introduction: Motor skills is a variable that marks the life of the human being. For this reason, predictive studies that explain the behavior of this variable are transcendental. Objective: The purpose of this study was to explore the main variables that predict motor skills according to classification tree analysis. Methods: A total of 291 Peruvian children aged 6 to 10 years (M=8.35; SD=1.29) participated in the study. They underwent a gross motor development test; a mathematics and reading test; a sociodemographic questionnaire; and body mass and height measurements. Results: The prediction results showed an initial model with 22 terminal nodes with 65.52% accuracy, and an optimized model with 10 terminal nodes with 68.34% accuracy. Discussion: This is the first study that applies machine learning by means of the classification tree model based on the CRISP-DM methodology to explore the main variables that predict motor skills in children aged 6 to 10 years. Conclusions: This study confirms that machine learning using classification tree modeling based on CRISP-DM methodology can predict motor skills in children aged 6 to 10 years with an accuracy of 68.34 %, with hours of physical activity practice per day being the most important variable, in addition to hours of screen device use per day and body mass of seven variables. © 2025 Federacion Espanola de Docentes de Educacion Fisica. All rights reserved.