One-way ANOVA.

This study addresses the research objective of predicting global happiness and identifying its key drivers. We propose a novel predictive framework that integrates unsupervised and supervised machine learning techniques to uncover the complex patterns underlying happiness scores across nations. Initially, we apply K-Means clustering to group countries based on similarities in their happiness patterns. For the first time, these cluster assignments are subsequently incorporated as additional features into ensemble learning models—specifically, Random Forests and XGBoost—to enhance the prediction of happiness scores. This hierarchical analysis approach yields a significant improvement in predictive performance, with an approximate 12% increase in R² compared to models that do not include clustering information. Using data from the World Happiness Report, our analysis reveals that global happiness can be categorized into three distinct groups (high, medium, and low). Among the various determinants examined, social support and GDP emerge as the most influential factors contributing to the happiness index. These findings not only advance the methodological framework for predicting happiness but also provide robust evidence for policymakers seeking to implement targeted interventions aimed at improving public well-being and promoting social progress.