🔐 URL Phishing Detection System

A complete full‑stack phishing URL detection app with machine learning backend and a user-friendly web GUI built using Flask, HTML, CSS, and JavaScript.

Live Demo

---

🔍 Overview

This application classifies URLs as Safe or Phishing by:

1. Extracting features from URLs (structural and content-based).
2. Training machine learning models (Random Forest, Logistic Regression, Naive Bayes).
3. Selecting Random Forest, which delivered ~96.9% accuracy.
4. Deploying via Flask, exposing a /predict API consumed by a JavaScript frontend.

---

💡 Key Features

• 🔬 ML-based detection: Random Forest performs best (~96.9% accuracy), compared to Logistic Regression (~93.4%) and Naive Bayes (~60.5%).
• 🧠 Feature engineering & EDA: Includes correlation heatmaps, pairplots, and pie‑charts for visualization.
• 🛠 GUI: Users can input a URL, trigger Flask backend, and view phishing predictions live.
• 🌐 Modern front-end stack: HTML/CSS for layout, vanilla JavaScript (or AJAX) for seamless interaction.
• ✅ Persistence: Trained model saved via pickle for deployment.

---

📈 Model Performance

Model	Accuracy	F1 Score	Recall	Precision
Random Forest	96.9%	97.2%	99.4%	98.9%
Logistic Regression	93.4%	94.1%	94.3%	92.7%
Naive Bayes	60.5%	45.4%	29.2%	99.7%

---

📦 Requirements

Flask
scikit-learn
pandas
numpy
matplotlib
seaborn

---

💻 Installation

Clone and install dependencies:

git clone https://github.com/garvitverma21/URL-Phishing-Detection-System.git
cd URL-Phishing-Detection-System
pip install -r requirements.txt

---

▶️ Running the App

1. Ensure random_forest_model.pkl is present:

   python phishing_detection.ipynb  # Or run training script

2. Start Flask:

   python app.py

3. Open http://localhost:5000 in your browser, enter a URL, and click Scan to see the result.