🚦 Chatdrive-ML

🔒 Malicious URL Detection API and ML Pipeline

🧪 Run the full notebook on Kaggle: Malicious URL ML (Kaggle)

Frontend Repo: chatdrive

🌐 Live Demo

Try the live application here: 🔗 ChatDrive

📝 Overview

This project provides a FastAPI-based web service and a machine learning pipeline for detecting potentially malicious URLs.

💬 It is designed to integrate with a chatting website, enabling real-time detection of malicious URLs shared in conversations.

It includes:

• 🔗 A REST API for URL analysis and prediction
• 🧩 Feature extraction and model inference using pandas and joblib
• 📓 A Jupyter notebook for model training and evaluation

🛠️ Tech Stack

• 🐍 Python: 3.11.13
• ⚡ FastAPI: Web API framework
• 📊 pandas, numpy: Data processing
• 📦 joblib: Model serialization
• 🤖 scikit-learn: ML pipeline (used in notebook)
• ✅ pydantic: Data validation
• 🌐 CORS: Enabled for API

⚙️ Installation

1. Clone the repository:

   git clone https://github.com/Satyam1Vishwakarma/chatdrive-ml.git
   cd chatdrive-ml

2. Install dependencies:

   pip install -r requirements.txt

🚀 Usage

▶️ Run the API server

uvicorn app:app --reload

The API will be available at http://localhost:8000.

🔗 API Endpoint

• POST /predict — Accepts a URL and returns a prediction (malicious or not).

📊 Model Training Offline

See malicious-url-ml.ipynb for the full ML workflow: data loading, feature engineering, model training, evaluation, and export.

📁 File Structure

• app.py — FastAPI app and feature extraction
• malicious-url-ml.ipynb — Jupyter notebook for ML pipeline
• static/ — Pretrained model files (url_model.pkl, etc.)
• requirements.txt — Python dependencies
• version.txt — Python version