AI Early-Warning System for Student Dropout Risk — Mohammed VI Polytechnic University and Morocco's Ministry of National Education

Researchers from Mohammed VI Polytechnic University (UM6P) in Benguerir, working with Morocco's Ministry of National Education, Preschool and Sports in Rabat, built a machine-learning system to flag students at risk of dropping out of school before they leave. The model was trained and tested on a large administrative dataset — about 1.4 million student profiles with more than 100 features, spanning five consecutive academic years — supplied by the Ministry for the Fès-Meknès region.

The team compared five algorithms (Decision Tree, Random Forest, XGBoost, LightGBM and an ensemble voting approach) and used SHAP (Shapley Additive exPlanations) to make each prediction interpretable to school administrators rather than a black box. Evaluated with a random data split, the best configuration reported 88% accuracy, 88% recall, 86% precision and an AUC of 87%; isolating the single best model (LightGBM, on Level 7 students) gave 83% accuracy and an 89% AUC.

The authors are explicit about a major caveat that changes how the headline numbers should be read: when they instead split the data chronologically — training on earlier years and testing on later ones, closer to how the system would actually be used in practice — performance dropped substantially, which the paper partly attributes to COVID-19-era disruption of normal dropout patterns. They also note the dropout class is a small minority of records, which complicates classification, and that the analysis, while large, covers only the Fès-Meknès region rather than all of Morocco.

As of publication (April 2025, posted to arXiv) this is a research system built on real ministry data, not yet reported as an operational early-warning tool embedded in schools' day-to-day guidance workflows.

Read the full analysis: https://www.researchgate.net/publication/390671139_AI-based_identification_and_support_of_at-risk_students_A_case_study_of_the_Moroccan_education_system