Power Plus (PowerPulse) - Electricity Demand Forecasting

Power Plus is a Flask web app that forecasts hourly electricity demand for six Delhi regions using a lightweight ensemble (XGBoost, LightGBM, Ridge). It pulls hourly weather from Open-Meteo and falls back to seasonal synthetic data if the API is unavailable.

Features

• 24-hour demand forecasts per selected region
• Ensemble predictions with a confidence score
• Weather-driven inputs with automatic fallback data
• Forecast plot generation saved to static/predicted_demand_plot.png
• JSON API for single-region predictions

Project layout

app.py                 Flask entry point
app/                   Optional blueprint-based app (not used by app.py)
data/                  Cleaned input datasets
saved_models/           Trained model artifacts (xgb, lgb, ridge)
templates/             HTML templates
static/                Images and generated plots
utils/                 Predictor, weather, and plotting utilities

Quick start

1. Install dependencies:

   pip install -r requirements.txt

2. Run the app:

   python app.py

3. Open: http://localhost:5000

Usage

• Forecast UI: http://localhost:5000/forecast
• Demo login (optional): username user, password password
• Map view: http://localhost:5000/map

API

POST http://localhost:5000/api/predict

Request body:

{
  "date": "2026-01-15",
  "region": "DELHI",
  "hour": 12,
  "temperature": 28.0,
  "humidity": 60.0,
  "wind_speed": 2.5
}

Response:

{
  "success": true,
  "prediction": {
    "ensemble": 3125.42,
    "xgb": 3150.12,
    "lgb": 3104.55,
    "ridge": 3008.77,
    "confidence": 0.94
  }
}

Notes

• Models are loaded from saved_models/. If any model files are missing, the app will fail to start.
• Weather data uses the Open-Meteo API and requires no API key.
• Update the SECRET_KEY in app.py before any real deployment.

Contact

• Suraj Mishra: iamsuraj187104@gmail.com
• GitHub: https://github.com/suraj187104/power-plus