Code Error Predictor + Fix Assistant

An intelligent debugging assistant that predicts programming errors and provides fixes with clear explanations using Machine Learning and Large Language Models.

---

Problem Statement

Debugging is one of the most time-consuming tasks in software development. Developers, especially beginners, often struggle to understand error messages and identify the correct fixes.

This results in:

• Increased development time
• Reduced productivity
• Difficulty in learning and understanding concepts

Traditional rule-based systems are not effective because error patterns are diverse, context-dependent, and continuously evolving. A scalable and adaptive solution is required.

---

Solution

This project builds a system that combines Machine Learning and LLMs to:

• Predict the type of error from code snippets or error logs
• Suggest possible fixes
• Provide clear, human-readable explanations

The system helps developers debug faster and understand errors more effectively.

---

Tech Stack

Backend

• Python
• FastAPI

Machine Learning

• scikit-learn
• Pandas
• NumPy
• TF-IDF Vectorization

LLM Integration

• OpenAI API / OpenRouter

Frontend

• React.js

---

Project Setup Instructions

Prerequisites

• Python Version: 3.13.x

1. Create the Virtual Environment

From the project root directory, run:

python -m venv venv

2. Activate the Environment

• Windows (PowerShell):

  .\venv\Scripts\Activate.ps1

• macOS/Linux:

  source venv/bin/activate

3. Install Dependencies

pip install -r requirements.txt

4. Run the Project

• Training and Evaluation: Run the main workflow to train the model and evaluate its performance:

  python -m src.main

• Prediction: Run a sample prediction using the trained model:

  python -m src.predict

---

How It Works

1. The user inputs an error message or code snippet.
2. The input is processed and converted into numerical features using TF-IDF.
3. A trained machine learning model predicts the error category.
4. The predicted error and input are sent to an LLM API.
5. The LLM generates:
   • Suggested fixes
   • Clear explanations
6. The results are displayed through a React-based user interface.

---

Features

• Error type prediction using machine learning
• Automatic fix suggestions
• Simple and understandable explanations
• Works with unseen error inputs
• FastAPI-based backend for efficient processing
• React-based user interface

---

Team and Responsibilities

Name	Responsibility
Abishek	Machine Learning model development, FastAPI backend integration
Lakshmi	Data collection, preprocessing, feature engineering
Lakshmi Shankar & Abishek	Frontend development using React, UI/UX design