The increasing prevalence of mobile phones in students’ lives has raised concerns regarding their impact on academic performance and well-being. This study examines both the positive and negative effects of mobile phone usage on students’ academic outcomes using a combined probabilistic and machine learning framework. Data were collected from 2500 college students across Tamil Nadu through a structured Likert-scale questionnaire designed to assess behavioral patterns, study habits, and mobile phone dependency. The questionnaire underwent validity and reliability testing using SPSS, including Cronbach’s alpha (α = [value]) for internal consistency and an exploratory factor analysis (EFA) for construct validation. A pilot study with 100 students was conducted before full-scale deployment. The data preprocessing pipeline included missing value imputation, normalization, feature selection, and bias analysis to ensure robust modeling. Probabilistic methods were applied to assess behavioral trends, while machine learning techniques such as K-Means Clustering, Principal Component Analysis (PCA), and Multiple Linear Regression were used to model academic performance outcomes. Models predicting positive academic outcomes (Q1–Q3) demonstrated high accuracy, with Model Q3 achieving 80.83% accuracy in classifying beneficial effects. Conversely, models predicting negative impacts (Q4–Q6) faced challenges, though Model Q4 attained 82.73% accuracy for non-negative classifications. Behavioral traits explained 82% of the variance in academic performance, with mobile phone addiction contributing to 35% of negative performance variance. These findings highlight the need for structured interventions such as screen-time management strategies and digital well-being initiatives to mitigate the adverse effects of excessive mobile phone use while leveraging its educational benefits. This study provides actionable insights for educators, policymakers, and researchers to enhance academic outcomes and minimize mobile phone-related distractions.