sekyikins/student-dropout-xgboost-shap-ghana

Machine learning system for predicting student dropout risk using an XGBoost classifier with SHAP-based explainability. Includes data preprocessing pipelines, model training scripts, evaluation experiments, and reproducibility configurations. It compares multiple baseline models and applies statistical testing to validate performance. # Student Dropout Prediction using XGBoost + SHAP ## Overview This project predicts student dropout risk using machine learning. The system is built using an XGBoost classifier and enhanced with SHAP to explain model decisions. ## Objectives - Predict students at risk of dropping out - Improve early academic intervention - Provide interpretable ML predictions for educators ## Methodology - Data preprocessing (missing value handling, encoding, feature cleaning) - Model training using XGBoost - Evaluation using cross-validation and hold-out test set - Model explanation using SHAP feature importance ## Models Used - Logistic Regression (baseline) - Random Forest - Support Vector Machine (SVM) - XGBoost (main model) ## Explainability SHAP is used to interpret feature contributions to dropout predictions. ## Reproducibility - Fixed random seed = 42 - Stratified train/validation/test split - Standard evaluation metrics: Accuracy, F1-score, ROC-AUC ## Tech Stack - Python - Scikit-learn - XGBoost - SHAP - Pandas, NumPy ## Author Student ML Research Project