lukanora/nigeria-education-analysis

Builds and compares three ML models (Logistic Regression, Decision Tree, Random Forest) to predict which secondary school students are at risk of dropping out. Best model achieves ROC-AUC of 0.873. Identifies distance to school as the single biggest driver of dropout. # 🎓 Predicting Secondary School Dropout Risk in Nigeria **Author:** Onyelukachukwu Nora Gwam **Tools:** Python · pandas · scikit-learn · matplotlib · seaborn **Data:** Simulated Nigerian secondary school student records (n = 850) **Best Model:** Random Forest (ROC-AUC = 0.873) --- ## 🎯 Project Overview Nigeria has one of the highest secondary school dropout rates in sub-Saharan Africa. Students who leave school early face significantly reduced lifetime earnings and limited access to opportunity. Early identification of at-risk students allows schools and policymakers to intervene before dropout occurs. This project builds and compares three machine learning classification models — Logistic Regression, Decision Tree, and Random Forest — to predict which students are at highest risk of dropping out, and identifies the structural factors that most strongly drive that risk. The analysis generates actionable, evidence-based policy recommendations for reducing dropout, particularly for girls in northern Nigeria. --- ## 🔍 Key Questions Answered 1. Which student characteristics most strongly predict dropout? 2. Does the gender gap in dropout risk vary by geopolitical zone? 3. Which machine learning model best identifies at-risk students? 4. What policy interventions would have the greatest impact on reducing dropout? --- ## 📊 Visualisations Produced | Figure | Description | |--------|-------------| | `fig1_dropout_by_zone_gender.png` | Dropout rate by zone and by gender within each zone | | `fig2_attendance_parent_edu.png` | Attendance distribution by dropout status + dropout by parent education | | `fig3_model_comparison.png` | ROC curves, metric comparison, and confusion matrix for best model | | `fig4_feature_importance.png` | Random Forest feature importance — the top drivers of dropout | --- ## 📁 Files ``` project2/ ├── nigeria_student_dropout.csv # Dataset (850 students × 13 features) ├── generate_data.py # Script used to generate t …