Student Dropout Risk Predictor (AI/ML Capstone Project)

An end-to-end machine learning system designed to identify students at risk of dropping out early, enabling timely academic interventions and counseling support.

📌 Problem Overview

High student dropout rates pose a significant challenge to educational institutions and student career outcomes. Identifying at-risk students before they fail or withdraw allows administrators, academic advisors, and parents to intervene early with targeted support.

This project implements a binary classification pipeline that analyzes academic performance, demographic factors, and enrollment trends to predict dropout probability and highlight top contributing risk factors.

📊 Dataset Description

The model is trained on the Predict Students' Dropout and Academic Success dataset, covering student records across multiple dimensions:

• Demographics & Social Factors: Age at enrollment, marital status, parents' qualification and occupation.
• Academic Performance: Curricular units enrolled, approved, evaluated, and semester grade averages (1st & 2nd semesters).
• Socio-Economic Indicators: Unemployment rate, inflation rate, and GDP.
• Target Variable: Binary status — 1 (Dropout / At Risk) vs. 0 (Graduate / Safe).

⚙️ Project Architecture & Pipeline

1. Data Preprocessing & Cleaning:

   • Filtered target outcomes to create a clear binary classification boundary (Dropout vs Graduate).
   • Removed target column leakage and missing value checks.
   • Applied StandardScaler to normalize numeric features across consistent scales.

2. Model Training:

   • Implemented an ensemble Random Forest Classifier (n_estimators=100, random_state=42) via scikit-learn.pipeline.Pipeline.
   • Used stratified 80/20 train-test splitting to maintain class proportions.

3. Evaluation & Explainability:

   • Evaluated using Accuracy, Precision, Recall, F1-Score, and Confusion Matrix.
   • Extracted tree-based feature importances to highlight the key drivers behind dropout risk.

📈 Results & Performance

• Overall Test Accuracy: 91.74%
• Evaluation Summary:
  • Robust classification on unseen test data.
  • High recall on the Dropout class, minimizing missed at-risk interventions.
  • Low false-positive rate, preventing unnecessary academic alerts.

Top Influential Risk Factors:

1. Curricular units 2nd sem (approved) — Course completion rate in the second semester.
2. Curricular units 1st sem (approved) — Early-stage academic success.
3. Curricular units 2nd sem (grade) — Academic grades achieved in semester 2.
4. Curricular units 1st sem (grade) — Baseline semester 1 grade point average.
5. Tuition fees up to date / Age at enrollment — Socio-economic and maturity factors.

🚀 How to Run the Project

1. Prerequisites

Install the required dependencies:

pip install pandas numpy scikit-learn matplotlib seaborn joblib