🏙️ Predicting Urban Building Electricity Consumption using Machine Learning

This project uses machine learning models (XGBoost, Random Forest, SVR) to predict electricity consumption of urban buildings based on the Seattle Building Energy Benchmarking 2016 dataset. It includes data preprocessing, model training, evaluation, and a fully interactive Flask dashboard.

---

📁 Project Structure

Predicting-Urban-Building-Electricity-Consumption/
├── app.py                      # Flask web application
├── main.py                    # Main ML pipeline controller
├── data/
│   └── 2016-building-energy-benchmarking.csv
├── models/
│   └── train_xgboost.py       # XGBoost training logic
├── preprocessing/
│   └── clean_data.py          # Data preprocessing pipeline
├── evaluation/
│   └── evaluate_models.py     # Model evaluation and plotting
├── outputs/
│   ├── X_train.csv, y_test.csv ...
│   ├── predictions_xgb.csv
│   ├── model_xgb.pkl
│   └── charts/
│       ├── feature_importance_xgb.png
│       ├── predicted_vs_actual_all_models.png
│       ├── model_comparison_metrics.png
│       └── residuals_analysis.png
├── templates/
│   ├── index.html
│   └── dashboard.html
└── static/ (optional)

---

🔧 How to Run

1. Clone the repository (or copy the project folder)

2. Create virtual environment (optional but recommended)

   python -m venv venv
   venv\Scripts\activate  # Windows

3. Install dependencies

   pip install -r requirements.txt

4. Run the ML Pipeline

   python main.py

5. Launch the Dashboard

   python app.py

6. Visit: http://localhost:5000/dashboard

---

📊 Features

Module	Description
✅ Preprocessing	Handles missing values, encoding, scaling, and feature selection
✅ XGBoost	Trained with both default and tuned parameters
✅ Evaluation	Calculates MAE, RMSE, R² + residual analysis and feature importance charts
✅ Dashboard	Interactive Flask web app with chart visualization and modal enlargement

---

📦 Dataset Used

Source: Kaggle - Seattle Building Energy Benchmarking 2016

• Total samples: ~3,300
• Features: Property type, GFA, EnergyStar score, electricity usage, etc.
• Target: SiteEnergyUse(kBtu)

---

📷 Website Preview

---

📜 License

This project is developed for academic and educational purposes under Multimedia University (MMU).