Enhance Student Well‑being and Digital Literacy with Machine Learning and Spatial Analysis

Advanced data analytics and machine learning can reshape the way schools cultivate both digital skills and student well‑being. We analysed data from three Italian high‑school classes (𝑁 = 64), combining random‑forest and neural‑network predictors with Spatial Autoregressive and Geographically Weighted Regression models. The approach captures how individual attributes, classroom geography and peer interactions jointly influence learning. Average grades rose from 5.34 to 6.15 and well‑being scores from 0.48 to 0.95 over one semester. Spatial estimates (𝜌 = 0.31, 𝑝 < 0.01) show that sitting next to high achievers yields a mean gain of 0.38 grade points, while local pockets of well‑being amplify the effect of digital‑literacy growth on performance. The results may support that digital‑literacy interventions, when delivered in spatially aware learning environments, produce measurable academic and affective benefits. The study offers a reproducible pipeline that blends machine‑learning prediction with spatial econometrics and provides evidence to guide data‑driven, equitable strategies for classroom design, teacher training and student support.