Modeling happiness in adults in academic training by applying machine learning algorithms.

Abstract

This study presents models of subjective well-being among Mexican college students using supervised classification algorithms, with the aim of identifying factors that predict happiness and evaluating the ability of different models to classify well-being levels. The study employs a quantitative, longitudinal, and explanatory design, utilizing BIARE microdata from 2019 to 2025. The sample included 6,793 observations and 52 attributes for university students aged 18 and older. The results show that the SimpleLogistic algorithm achieved the best performance, with an accuracy of 99.72%. Positive factors associated with happiness include monthly solvency, financial security, present and annual well-being, purpose, and daily achievement.