Built a machine learning system using XGBoost to predict student depression risk, featuring real-time Streamlit dashboards and SHAP model explainability.
Key Technical Features
Achieved 85% accuracy using XGBoost on 10K+ records
Built real-time dashboards using Streamlit and integrated SHAP for explainability
Reduced report generation time by 30%, improved system robustness via debugging
Collaborated in Agile sprints and resolved data inconsistencies