Bax-dev/AI-Power-Outage-Predictor

AI Power Outage Predictor for energy loss and gain in Nigeria # AI Power Outage Predictor A completely AI-powered system that predicts power outage probability using Random Forest machine learning. The system analyzes time patterns, day of week, location, and environmental factors to provide accurate outage predictions. ## Features - **Fully AI-Powered**: Uses Random Forest classifier for intelligent predictions - **Web Interface**: Beautiful, modern HTML homepage with interactive predictions - **Feature Engineering**: Advanced time-based features (cyclic encoding, peak hours, seasons) - **Location-Aware**: Supports multiple locations with location-specific patterns - **Time-Based Analysis**: Considers hour of day, day of week, and seasonal patterns - **Probability Output**: Provides percentage probability of power outage - **Risk Assessment**: Categorizes risk levels (Low, Moderate, High, Very High) ## Technology Stack - **Python 3.8+** - **Flask**: Web framework for HTML interface - **Pandas**: Data manipulation and analysis - **Scikit-learn**: Random Forest machine learning model - **NumPy**: Numerical computations - **Joblib**: Model persistence ## Installation 1. Clone or download this repository 2. Install required packages: ```bash pip install -r requirements.txt ``` ## Quick Start ### Step 1: Generate Historical Data Generate synthetic historical power outage data for training: ```bash python data_generator.py ``` This creates `historical_outage_data.csv` with 365 days of hourly data. ### Step 2: Train the AI Model Train the Random Forest model on the historical data: ```bash python train_model.py ``` This will: - Engineer features from raw data - Train a Random Forest classifier - Evaluate model performance - Save the trained model as `power_outage_model.pkl` ### Step 3: Make Predictions #### Option A: Web Interface (Recommended) Start the web server: ```bash python app.py ``` Then open your browser and navigate to: ``` localhost ``` You'll see a beautiful homepage where you ca …