Mansour-Essgaer/Predicting-High-School-Dropout-Vulnerability-in-South-Libya

Armed conflict and institutional fragility have severely destabilized the educational landscape in Libya, particularly in the southern Fezzan region # 🎓 An Interpretable Machine Learning Framework for Predicting High School Dropout Vulnerability in South Libya > **Official Repository** for the paper: *"Explainable Ensemble Learning for Student Dropout Prediction in Conflict-Affected Educational Systems"* > Published in: *2026 IEEE 5th International Maghreb Meeting of the Conference on Sciences and Techniques of Automatic Control and Computer Engineering (MI-STA)*, 2026. > DOI: [10.1109/MI-STA68962.2026.11511168] ## 📌 Overview Armed conflict and institutional fragility have severely destabilized the educational landscape in Libya, particularly in the southern Fezzan region (e.g., Sabha). This repository accompanies our research that frames high school student dropout vulnerability as a binary classification task, distinguishing between students likely to pursue **Continued Enrollment** and those at high risk of **Dropping Out**. Moving beyond black-box predictions, this study introduces a rigorous, interpretable Educational Data Mining (EDM) framework. We engineered a novel dataset, constructed a synthetic target variable via a rule-based scoring mechanism (strictly separated from predictive features to prevent data leakage), and benchmarked 12 machine learning algorithms. Our optimal **Stacking Classifier** achieved a robust F1-score of **0.924**. Furthermore, we employed **LIME** for local explainability and **K-Means clustering** for dual-layered risk profiling, providing actionable, granular insights for educational policymakers. --- ## 📊 Dataset Description The dataset comprises primary survey data collected from **1,158 high school students** across five schools in Sabha, Libya, during the 2023 academic year. The 34-item structured questionnaire was adapted to capture the unique socio-cultural, psychological, and academic nuances of the Libyan context. ### 📈 Corpus Statistics & Attributes | Attribute Category | Key Features Included | Data Types | | :--- | :--- | :--- | | **Demographic** | Gender …