Student Grade Prediction

This project predicts student grades (G3) based on various features such as parental education, study time, number of failures, and more. Machine learning models including Linear Regression, K-Nearest Neighbors (KNN), and Random Forest are used for making predictions.

Project Structure

The project is organized into separate Python files, each responsible for a specific task. The key components of the project are:

Files

• preprocess.py: Contains functions for loading, cleaning, and scaling the dataset.
• log_regr.py: Implements the Linear Regression model, trains it, and evaluates its performance.
• knn_model.py: Implements the K-Nearest Neighbors (KNN) model, trains it, and evaluates its performance.
• random_forest.py: Implements the Random Forest model, trains it, and evaluates its performance.
• evaluation.py: Compares the performance of all three models and prints evaluation metrics (MSE, R²).
• visualize.py: Provides functions for visualizing the results, including correlation heatmaps and model performance comparison.

Project Workflow

1. Preprocessing:

   • The preprocess.py file handles the loading and preprocessing of the dataset. It cleans the data by selecting only relevant columns and scales the features using StandardScaler.
   • The columns used in the model are:
     • Medu, Fedu, goout, Walc, failures, studytime, absences, G1, G2, and G3 (with G3 as the target).

2. Model Training:

   • Models are defined in separate files (log_regr.py, knn_model.py, random_forest.py). Each file:
     • Loads the preprocessed data from preprocess.py.
     • Trains a corresponding machine learning model.
     • Evaluates the model's performance using Mean Squared Error (MSE) and R².

3. Model Evaluation:

   • The evaluation.py file compares the performance of all three models (Linear Regression, KNN, Random Forest) and prints out their MSE and R².

4. Visualization:

   • The visualize.py file provides functions to visualize model performance and feature correlations:
     • A heatmap of feature correlations is plotted to show relationships between features and the target.
     • A bar chart is used to compare the performance of the models (MSE and R²).

---

Setup and Installation

Prerequisites

To run this project, you'll need to have Python installed along with the following libraries:

• pandas
• numpy
• scikit-learn
• matplotlib
• seaborn

You can install the required dependencies using pip. Create a virtual environment and run the following command:

pip install pandas numpy scikit-learn matplotlib seaborn