Getting a full picture of student proficiency and well-being in an online or blended learning environment can be difficult. Among the traditional methods that usually overlook students' everyday activities, social media usage, or degrees of stress are standardized tests or subjective assessments. We analyzed these patterns using data from over 1,000 students from diverse backgrounds. The data contains specifics such as time spent in online classes, online experience ratings, everyday activities (self-study, exercise, social networks, TV), changes in health, and even stress-releasing techniques.

We seek to identify trends using machine learning (ML) techniques otherwise hard to reveal. For example, using clustering techniques we identified groups of students displaying similar behavior including stress management or balance between leisure and study time. Our analysis revealed two student profiles: an "Academically-Engaged" group (68% of students) distinguished by more study time and less social media usage, and a "Leisure-Focused" group (32%) showing reverse trends. Principal Component Analysis (PCA), whose first two components accounted for 69% of the variance in student behavior, further confirmed these groups.

Such an ML-driven study draws attention to close links among general well-being, online activity, and study practices. The study found clear links between activity patterns and learning outcomes as academically engaged students reported superior satisfaction levels (2.74/5.0 vs 2.43/5.0, p=0.001). Weight management also clearly differed between groups as academically engaged students kept more stable weight patterns (47.5% vs 38.3% reporting constant weight, p=0.035). Interestingly, both groups showed comparable patterns in family ties and health issues, suggesting that additional factors outside of study conduct could be involved.

Through clustering and correlation analyses, this work exposes the relationships among elements influencing student performance. Our results show how everyday activities, time management, and lifestyle choices collectively affect academic satisfaction and general well-being, therefore transcending the dependence on grades or standardized tests. The diverse profiles found by machine learning suggest that support systems should be customized to fit various student behavioral patterns, thus acknowledging the close relationship between academic engagement and personal wellness. Through this data-driven approach, educators can create more focused interventions incorporating both academic and lifestyle elements, thereby enabling students to achieve better learning outcomes while maintaining their physical and mental well-being.

Keywords: Education, Educational Technology, Machine Learning, Student Assessment, Skills and Well-being Evaluation.