🎓 Student Performance Prediction with Linear Regression

📝 Short Description

This project analyzes and predicts student academic performance using a linear regression model on the Student Performance Dataset (student-mat.csv).
It includes data preprocessing, exploratory data analysis (EDA), model training, evaluation, and prediction for new student data.

---

🚀 Features

• Data Preprocessing:
  • Encodes categorical variables into numeric form.
  • Checks and handles missing data.
• Exploratory Data Analysis:
  • Histograms, boxplots, and lineplots for grade distribution and relationships.
  • Correlation heatmap of numerical features.
• Modeling:
  • Splits data into training/testing sets.
  • Trains a LinearRegression model using scikit-learn.
  • Evaluates performance with MSE and R² score.
• Prediction:
  • Predicts grades for hypothetical students.
• Model Saving:
  • Exports trained model with joblib for future use.

---

📦 Requirements

Install dependencies:

pip install pandas matplotlib seaborn scikit-learn joblib

---

📂 Dataset

The dataset contains demographic, social, and academic attributes for students.
Target variable: G3 — the final grade.

---

⚙️ Usage

1. Place student-mat.csv in the project directory.
2. Run the notebook or script to:
   • Explore data
   • Train the model
   • Make predictions
3. Example prediction:

prediction = model.predict(new_data)
print(f'Predicted final grade: {prediction[0]:.1f}')

4. The trained model is saved as new_model.joblib.

---

📊 Outputs

• Visualizations:
  • Grade distribution histogram
  • Boxplots (by school, gender, alcohol consumption, study time, etc.)
  • Correlation heatmap
• Metrics:
  • Mean Squared Error (MSE)
  • R² score
• Saved Model:
  • File: new_model.joblib

---

🔍 Workflow

1. Data Loading — Load student-mat.csv into a Pandas DataFrame and inspect structure.
2. Preprocessing — Encode categorical variables into integers and prepare X (features) & y (target).
3. EDA — Use Seaborn/Matplotlib for distribution plots, boxplots, and correlation analysis.
4. Model Training — Split data (80% train, 20% test), fit a LinearRegression model.
5. Evaluation — Predict on the test set and compute MSE & R² score.
6. Prediction — Create new student profile → predict final grade.
7. Model Saving — Save trained model with joblib.dump() for reuse.

---

Author: Nimona Engida
Dataset Source: UCI Machine Learning Repository — Student Performance