---
title: Intrusion Detection Dashboard
emoji: 🛡️
colorFrom: indigo
colorTo: blue
sdk: streamlit
sdk_version: "1.31.1"
app_file: app.py
pinned: false
---

This is a machine learning Streamlit app that predicts potential cyberattacks based on real-time session characteristics like IP reputation, login attempts, and encryption type.

It uses a LightGBM classifier trained on a labeled intrusion detection dataset. The model prioritizes **recall** to minimize undetected attacks and is deployed via a Hugging Face API.

- 📊 Explore session data trends
- 🔍 Predict intrusions in real time
- 🤖 Model: LightGBM with threshold = 0.2

[🔗 Model Notebook](https://github.com/butlerem/intrusion-detection-model-lgbm)  
[🔗 Dataset Source](https://www.kaggle.com/code/nukimayasari/cybersecurity-intrusion)