🎓 Student Performance Prediction System

Course: CIE 317/417 Machine Learning
Institution: Zewail City of Science, Technology and Innovation

---

📌 Project Overview

A comprehensive Machine Learning system that analyzes educational data and predicts student performance. Leveraging a dataset of 20,000+ student records, the system identifies key factors influencing academic success and provides predictive insights through three core tasks:

1. Regression: Predicting exact final_score (0-100)
2. Binary Classification: Determining pass_fail status
3. Multiclass Classification: Predicting specific final_grade (A, B, C, D, F)

An interactive Streamlit Dashboard visualizes model performance and enables real-time predictions.

---

👥 Team Members

Name	Student ID
Mohammed Ali Sadek	202200594
Ahmed Amgad	202200393
Abdulrahman Madgy	202200341
SalahDin Ahmed Rezk	202201079

---

📂 Dataset Details

Source: Term_Project_Dataset_20K.csv

• Size: 20,000+ samples
• Features: 40 input variables across 4 categories
• Target Variables: final_score, final_grade, pass_fail

Feature Categories

Category	Examples
Demographic	Age, Gender, Parent Income, Sibling Count
Academic History	Previous GPA, High School Grade, Attendance Rate
Behavioral	Study Hours, Participation, Alcohol Consumption
Psychological	Stress Level, Motivation, Anxiety, Sleep Hours

---

⚙️ Methodology & Pipeline

1. Exploratory Data Analysis (EDA)

• Distribution analysis of grades
• Correlation matrices identifying relationships (e.g., Study Time vs. Score)
• Outlier detection and visualization

2. Data Preprocessing

• Imputation: Handling missing values in numerical and categorical columns
• Encoding: One-Hot Encoding for nominal features (e.g., Gender)
• Balancing: SMOTE (Synthetic Minority Over-sampling Technique) for class imbalance

3. Model Training

• Training multiple classical ML models
• Hyperparameter tuning via GridSearchCV/RandomizedSearchCV
• Cross-validation for robust performance estimation

4. Evaluation

• Regression Metrics: RMSE, MAE, R² Score
• Classification Metrics: Accuracy, Precision, Recall, F1-Score
• Visualizations: Confusion Matrices, ROC Curves, Feature Importance

---

🧠 Models Implemented

📉 Regression Models (Score Prediction)

• Linear Regression
• Ridge & Lasso Regression
• Random Forest Regressor
• Support Vector Regressor (SVR)
• Gradient Boosting Regressor

📊 Classification Models (Grade/Pass-Fail Prediction)

• Logistic Regression
• Random Forest Classifier
• Gradient Boosting Classifier
• XGBoost Classifier
• Support Vector Machine (SVM)
• K-Nearest Neighbors (KNN)

---

💻 Installation & Usage

Prerequisites

• Python 3.8 or higher
• pip package manager

1. Clone the Repository

git clone https://github.com/aboalis/student-performance-prediction.git
cd student-performance-prediction

2. Install Dependencies

pip install -r requirements.txt

3. Run the Jupyter Notebook

To view the complete analysis and training process:

jupyter notebook notebooks/Final_Project.ipynb

4. Launch the Streamlit Dashboard (Optional)

For interactive predictions and visualizations:

streamlit run app.py

The dashboard will open in your browser at http://localhost:8501

---

📁 Repository Structure

student-performance-prediction/
│
├── data/
│   └── Term_Project_Dataset_20K.csv   # Primary dataset
│
├── notebooks/
│   └── Final_Project.ipynb            # Main analysis & training notebook
│
├── models/                            # Saved trained models (generated)
│
├── app.py                             # Streamlit dashboard application
├── requirements.txt                   # Python dependencies
├── README.md                          # Project documentation
└── .gitignore                         # Git ignore file

---

📊 Key Findings

Top Predictive Features

1. Previous GPA - Strongest predictor of academic success
2. Study Hours per Week - Strong positive correlation with final scores
3. Attendance Rate - Critical factor in pass/fail outcomes
4. Sleep Hours - Significant impact on cognitive performance

Model Performance Highlights

• Best Regression Model: Random Forest Regressor
• Best Binary Classifier: XGBoost
• Best Multiclass Classifier: Gradient Boosting

Insights

• Non-linear relationships between features favor ensemble methods (Random Forest, XGBoost)
• SMOTE balancing significantly improved minority class predictions (F grades)
• Behavioral factors (study time, participation) outweigh demographic factors in importance

---

🚀 Future Enhancements

• Deep Learning models (Neural Networks) for comparison
• Feature engineering with polynomial features
• Real-time data integration with student information systems
• Mobile application deployment
• Explainable AI (SHAP values) for model interpretability

---

📜 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

🙏 Acknowledgments

Course: CIE 317/417 Machine Learning
Instructor: Dr. Ahmed Tolba
Institution: Zewail City of Science, Technology and Innovation

Tools & Libraries:

• Python, Scikit-Learn, XGBoost
• Pandas, NumPy, Matplotlib, Seaborn
• Streamlit, Jupyter Notebook
• Google Colab

---

👤 Contact

Mohammed Ali Sadek

---

Project Date: Fall 2024
Last Updated: January 2025