Student GPA Prediction: Linear Regression from Scratch

Project Overview

This repository contains a hands-on machine learning project focused on predicting student GPAs. Unlike using pre-built libraries for the core algorithms, I've built the entire linear regression model from the ground up using fundamental Python libraries. The project walks through a complete machine learning workflow, from initial data cleaning to final model evaluation.

This project was a great way for me to get a deep understanding of the mechanics behind a machine learning model.

Key Features:

• From-Scratch Implementation: Custom Python functions for prediction, the cost function (MSE), and gradient descent.
• Custom Data Preprocessing: A self-made MyStandardScaler class is used to standardize the data.
• Complete Workflow: The project script covers data loading, cleaning, exploratory data analysis (EDA) with plots, model training, evaluation, and visualization of results.
• Visual Analysis: Histograms, bar charts, and a final scatter plot are used to analyze data distributions and model performance.

What's Included

student-performance-predictor/
├── Student_gpa_prediction.py    # The main Python script
├── Student_performance_data.csv   # The dataset used for this project
├── Student_performance_data.pdf   # Additional project documentation
└── README.md                      # This documentation

Getting Started

To run this project, follow these simple steps.

1. Clone the repository

First, get a local copy of the project files.

git clone https://github.com/yassinexng/student-performance-predictor.git
cd student-performance-predictor

2. Install required packages

You'll need a few common data science libraries.

pip install pandas matplotlib scikit-learn

Important: You may encounter a compatibility issue with NumPy 2.x and other libraries. If you get an ImportError related to NumPy, please use the following command to install a compatible version:

pip install "numpy<2"

3. Run the script

Finally, execute the Python script from your terminal.

python mainfunction.py

The script will print its progress to the console and generate several plots.

Implementation Details

Libraries Used

• NumPy: For numerical operations, array manipulation, and the core linear regression algorithm implementation.
• Pandas: For efficient data loading, cleaning, and manipulation of the dataset.
• Matplotlib: For creating all the visualizations, including histograms, bar charts, and the final prediction vs. actual GPA plot.
• scikit-learn: Used only for splitting the data into training and testing sets, keeping the core model implementation completely custom.

Dataset

The included Student_performance_data.csv contains a synthetic dataset with student information, including:

• StudyTimeWeekly
• Absences
• ParentalEducation
• Extracurricular participation
• GPA values (the target variable for prediction) It can be found within the same repo of this project.

About This Project

As my first machine learning implementation, this project represents an important step in my journey. I've focused on:

• Clear, documented code

• Thorough explanations of the math behind the algorithms

• Practical data preprocessing techniques

• Visual interpretation of results

I welcome any feedback or suggestions for improvement - please feel free to open an issue or reach out directly in my LinkedIn. This project was my first attempt at building a complete machine learning solution from scratch. My focus was on creating clear, well-documented code that explains the fundamental principles of linear regression, from the math behind the cost function to the iterative process of gradient descent. This was a valuable learning experience, especially in overcoming common challenges like library compatibility issues. I hope this code serves as a useful and clear example for others starting their own journey in data science.