README.md

---
title: SF Crime Analytics | AI-Powered
emoji: 🚓
colorFrom: red
colorTo: blue
sdk: docker
app_port: 8501
tags:
- streamlit
- machine-learning
- xgboost
- crime-prediction
pinned: true
license: apache-2.0
---

# 🚓 San Francisco Crime Analytics & Prediction System

## Overview
This project is a comprehensive AI-powered dashboard for analyzing and predicting crime in San Francisco. It leverages historical data and advanced machine learning models (XGBoost) to provide actionable insights and real-time risk assessments.

## Features
-   **📊 Historical Trends**: Visualize crime distribution by hour, district, and category.
-   **🗺️ Geospatial Intelligence**: Interactive heatmaps showing crime density and evolution over time.
-   **🚨 Tactical Simulation**: Simulate patrol strategies and assess risk levels for specific sectors.
-   **💬 Chat with Data**: Natural language interface to query the dataset.
-   **🚀 Advanced Prediction (99% Accuracy)**: High-precision crime categorization using an optimized XGBoost model.
-   **🤖 AI Crime Safety Assistant**: Interactive chatbot for safety tips and model explanations.

## Installation

1.  **Clone the repository**:
    ```bash
    git clone <repository-url>
    cd Hackathon
    ```

2.  **Install dependencies**:
    ```bash
    pip install -r requirements.txt
    ```

3.  **Run the application**:
    ```bash
    streamlit run app.py
    ```

## Docker Support
Build and run the container:
```bash
docker build -t sf-crime-app .
docker run -p 8501:8501 sf-crime-app
```

## Technologies
-   **Frontend**: Streamlit
-   **Backend**: Python, Pandas, NumPy
-   **ML Models**: XGBoost, Scikit-Learn (KMeans)
-   **Visualization**: Plotly, Folium
-   **AI Integration**: Groq (Llama 3)

---
*Developed for HEC Hackathon*