⚡ Energy Demand Forecaster

A machine learning project that forecasts electricity demand using time-series energy market data. The system applies feature engineering techniques such as lag features, rolling statistics, and time-based variables, and uses an XGBoost regression model for short-term demand prediction. It is designed for infrastructure planning and energy resource optimization.

---

📌 Project Objectives

• Forecast electricity demand using historical energy market data
• Engineer meaningful time-series features (lags, rolling averages, time features)
• Train a high-performance XGBoost regression model
• Evaluate model performance using regression metrics
• Visualize predictions and insights for analysis

---

📁 Folder Structure

Energy_demand/
│
├── data/
│ └── electricity_demand.csv
│
├── models/
│ └── xgboost_forecaster.pkl
│
├── reports/
│ ├── forecast_results.csv
│ ├── model_metrics.csv
│ └── figures/
│ ├── actual_vs_forecast.png
│ ├── feature_importance.png
│ ├── residuals.png
│ └── demand_trend.png
│
├── src/
│ ├── data_loader.py
│ ├── feature_engineering.py
│ ├── feature_pipeline.py
│ ├── model.py
│ ├── model_pipeline.py
│ ├── evaluation.py
│ ├── visualization.py
│ └── pipeline.py
│
├── tests/
│ ├── test_data_loader.py
│ ├── test_features.py
│ └── test_model.py
│
├── dashboard.py
├── main.py
├── requirements.txt
└── README.md 

---

⚙️ Installation Guide

1. Clone repository

git clone https://github.com/your-username/infrastructure-demand-forecaster.git
cd infrastructure-demand-forecaster

2. Create virtual environment

python -m venv venv

3. Activate Virtual Environment

Windows

venv\Scripts\activate

Mac/Linux

source venv/bin/activate

4. Install dependencies

pip install -r requirements.txt

🚀 How to Run

Step 1: Train model

python main.py

Step 2: Run dashboard

streamlit run dashboard.py

📊 Results & Visualizations

📈 Actual vs Forecast

📉 Feature Importance

📊 Residual Analysis

📈 Demand Trend

📏 Model Performance

Metric	Score
MAE	41.66
RMSE	52.19
R²	0.897

🧠 Techniques Used

• Time-series feature engineering
• Lag features
• Rolling statistics
• XGBoost regression
• Train/test split
• Data visualization

📌 Technologies

• Python
• Pandas
• NumPy
• Scikit-learn
• XGBoost
• Streamlit
• Plotly

👨‍💻 Author

Name: MD Moshiur Rahman

📄 License

MIT License