Student Retention Predictive Model
Predictive model to identify at-risk students and improve retention rates
Situation
High dropout rates in engineering programs were affecting program reputation and student outcomes.
Task
Develop predictive model to identify at-risk students early and enable timely interventions.
Action
Business & Strategy
Analyzed survey and academic data to identify behavioral patterns correlated with dropout risk. Collaborated with faculty to design intervention strategies based on model predictions.
Technical Implementation
Trained Random Forest classifier on historical student data (5 years, 3,000+ students). Built Flask dashboard for faculty to view at-risk students and track intervention effectiveness. Implemented feature importance analysis to identify key risk factors.
Results
Business Impact
Retention increased by 15% post-intervention. Enabled proactive support for struggling students, improving overall program outcomes.
Technical Achievement
Model AUC: 0.88. Successfully deployed on school intranet with 99.5% uptime. Feature importance analysis identified top 5 risk indicators.