---
title: NSL-KDD Anomaly Detection
emoji: 🧠
colorFrom: blue
colorTo: green
sdk: streamlit
app_port: 8501
pinned: false
tags:
  - anomaly-detection
  - machine-learning
  - streamlit
  - cybersecurity
---

# 🧠 NSL-KDD Anomaly Detection (Isolation Forest vs One-Class SVM vs Logistic Regression)

This Streamlit web app performs **network intrusion detection** using the **NSL-KDD dataset**,  
comparing both **unsupervised** and **supervised** machine learning algorithms.

---

## 🌐 Live Demo
👉 Try it on **Hugging Face Spaces**:  
(Replace this link with your Space URL after deployment)

---

## 🧠 Overview
The app demonstrates anomaly detection for cybersecurity by comparing three models:

| Model | Type | Description |
|--------|------|-------------|
| **Isolation Forest** | Unsupervised | Detects anomalies by isolating data points |
| **One-Class SVM** | Unsupervised | Learns a decision boundary around normal data |
| **Logistic Regression** | Supervised | Classifies known attacks using labeled data |

The dataset comes from the **NSL-KDD** benchmark — an improved version of the KDD Cup '99 dataset.

---

## ⚙️ Workflow
1. Load the NSL-KDD dataset from an online source  
2. One-hot encode categorical columns (`protocol_type`, `service`, `flag`)  
3. Scale all numeric features with `StandardScaler`  
4. Train and compare all three models  
5. Display metrics and confusion matrices in Streamlit

---

## 📊 Example Outputs
- **Model Performance Table** – Accuracy, Precision, Recall, F1-score  
- **Confusion Matrices** – For all three models  
- **Visual Insights** – Easy comparison between supervised and unsupervised models  

---

## 🧾 Dataset
Source: [University of New Brunswick – NSL-KDD Dataset](https://www.unb.ca/cic/datasets/nsl.html)

It contains 41 features describing network traffic and a label (`normal` or `attack`).

---

## 🧰 Tech Stack
- **Python 3.9+**
- **Streamlit** – Web framework  
- **Scikit-learn** – ML algorithms & preprocessing  
- **Matplotlib / Seaborn** – Visualization  
- **Pandas / NumPy** – Data handling  

---

## 🏗️ Local Setup (Optional)
```bash
# Clone the repo
git clone https://huggingface.co/spaces/your-username/nsl-kdd-anomaly-detection
cd nsl-kdd-anomaly-detection

# Install dependencies
pip install -r requirements.txt

# Run Streamlit app
streamlit run app.py