Student Performance Prediction

Overview

This project analyzes various factors that influence student academic performance using machine learning classification algorithms. It provides data visualization and predictive modeling to identify key factors that impact student grades.

Project Structure

• Project.py - Main application with interactive data visualization and ML model training
• TestData.csv - Student performance dataset

Current State

• ✅ Data visualization capabilities functional
• ✅ All ML models (Decision Tree, Random Forest, Perceptron, Logistic Regression, MLP) working
• ✅ Custom student testing functional

Features

1. Interactive Data Visualization: 9 different graph types showing student performance patterns
2. Multiple ML Models: Compares 5 different classification algorithms
3. Performance Metrics: Accuracy, precision, recall, F1-score for each model
4. Custom Prediction: Allows input of student characteristics for performance prediction

How to Use

The application runs interactively through the console workflow. Users can:

1. Choose from 9 visualization options (1-9) or exit (10)
2. View accuracy metrics for all ML models
3. Optionally test custom student input for predictions

Technical Details

• Language: Python (3.7-3.11)
• Key Libraries: pandas, scikit-learn, matplotlib, seaborn, numpy
• Data: 481 student records with 17 features
• ML Models: Decision Tree, Random Forest, Perceptron, Logistic Regression, MLP Classifier

Architecture

This is a console-based machine learning application that:

1. Loads student data from CSV
2. Provides interactive data visualization options
3. Trains multiple ML classification models
4. Evaluates model performance with detailed metrics
5. Offers custom prediction capabilities

The application follows a linear workflow: data loading → visualization options → model training → evaluation → optional custom prediction.

Expected outcomes

Typical accuracy ranges you can expect from this dataset and these models:

1. Decision Tree: ~75–85% (interpretable, may overfit)
2. Random Forest: ~80–90% (usually best overall)
3. Perceptron: ~70–80% (linear baseline)
4. Logistic Regression: ~75–85% (probabilistic predictions)
5. Neural Network (MLP): ~80–88% (captures non-linear patterns)

In general, engagement metrics (raised hands, visited resources, discussions) and attendance are the strongest predictors.

Practical Applications

This tool can help:

1. Teachers— identify at-risk students early
2. Administrators— allocate tutoring/resources better
3. Parents— understand what affects performance
4. Students— see how engagement impacts their grades
5. Researchers— study educational patterns

Future Goal

Is to determine factors common in successfull people and give a report to individuals to improve in areas like Emotional Intelligence, Pattern learning, Risk Capacity