Exploring the Impact of Psychological Needs on Physical Activity Using a Logistic Regression-Based Machine Learning Model

Physical activity (PA) plays a vital role in maintaining overall health and well-being. The current study presents a logistic regression model developed to predict the PA level of young adults based on psychological variables. The PA of 422 subjects were recorded and a k-means clustering technique was used to group the participants’ level of PA while their responses on the psychological need for exercise satisfaction were utilized to ascertain the influence of the psychological variables on their PA and develop a predictive model using logistic regression (LR). The model's performance was evaluated using cross-validation, with unseen data reserved for testing and validation. The results revealed exceptional predictive capabilities, with high accuracy scores achieved in all stages of model development ranging from 98 to 99%. A confusion matrix analysis highlighted minimal misclassifications, further confirming the model's effectiveness. In the training stage, only one misclassification occurred, while both the testing and validation stages showed a single misclassification where a high PA level individual was misclassified as a low PA level. These findings underscore the model's robustness in accurately differentiating between high and low PA levels. These results demonstrate the potential of the logistic regression model, supported by the utilization of cross-validation, as a reliable tool for predicting PA status based on psychological variables in young adults.