Deanthestallion/student-dropout-risk-predictor

Early-warning ML tool flagging secondary-school students at risk of dropping out, from attendance and grades. 3MTT capstone, Nigeria. RandomForest, 85% recall on the at-risk class, SHAP explanations, Streamlit app. # Student Dropout Risk Predictor for Secondary Schools **3MTT Capstone Project | Nigeria** A machine learning early-warning tool that scores a secondary-school student's risk of finishing the year below the pass mark, places them in a Low / Medium / High band, and shows the teacher exactly which factors drove that score. Every number in this README was produced by running notebook.ipynb. Nothing is estimated or copied from a paper. The notebook in this repository is saved with its outputs, so each figure can be traced to the cell that produced it. ``` notebook.ipynb full pipeline, executed, with outputs and 44 passing assertions app.py Streamlit app for teachers models/ saved pipeline (preprocessing + SMOTE + RandomForest) via joblib evaluation/ model_comparison.csv, metrics.json, and 14 plots data/ UCI CSVs plus download_data.py scripts/ build_notebook.py, and the verification scripts below requirements.txt pinned versions ``` Three checks can be run at any time to confirm the repository is internally consistent: ```bash python scripts/verify_repo.py # files present, notebook clean, docs match metrics.json, # leakage guarantees hold, no em dashes python scripts/verify_notebook.py # every cell executed, every assertion passed python scripts/test_app.py # drives app.py through Streamlit's test harness and # asserts each sample profile renders a band plus factors ``` `verify_repo.py` re-reads `evaluation/metrics.json` and asserts that each headline number quoted in this README matches it, so the documentation cannot silently drift away from the results. --- ## Problem Nigeria carries one of the largest out-of-school populations in the world. UNESCO and UNICEF estimates put the figure at roughly 10 to 20 million children, and the transition from junior to senior secondary is one of the points where students are most likely to leave. The drivers are well doc …