Bengaluru Renewable Energy Forecasting System

Project Overview

This project builds a location-agnostic solar forecasting and suitability mapping system for Bengaluru.
It predicts daily solar energy output using NASA POWER data and machine learning models (XGBoost, Ridge), while also mapping zone-level suitability for solar installations.

The final deliverable is a Streamlit dashboard that allows users to:

• View solar suitability across Bengaluru zones.
• Check the solar potential of their own home by entering coordinates.
• Download full zone-level datasets for further analysis.

---

Features in Streamlit App

• Interactive Suitability Map
  Displays Bengaluru divided into 223 zones, colored by solar suitability (score 0–100 or annual kWh).
• Top 10 Zones
  Automatically ranks the best-performing zones by score or annual kWh.
• Home Check Tool
  Lets users check solar feasibility at their location (manual coordinates or geolocation).
• Download Data
  Option to download the full CSV file of all zone-level predictions.

---

Model Accuracy (XGBoost Example)

Below is an example for Zone BLR-0001 showing Actual vs Predicted Daily Energy using the XGBoost model on the test set:

---

Implementation Steps (Phase 1 → Phase 7)

Phase 1 — Data Collection

• Fetch daily solar irradiance and climate variables from NASA POWER API.
• Store raw CSV/Parquet files in /data/raw/.

Phase 2 — Data Preprocessing

• Clean missing values, normalize metrics.
• Engineer features (temperature, irradiance, wind speed).
• Store processed files in /data/processed/.

Phase 3 — Feature Engineering

• Create lag features (moving averages of irradiance/temperature).
• Encode seasonal/time-of-year effects.
• Split into train/test sets.

Phase 4 — Model Training

• Train Ridge Regression and XGBoost Regressor models.
• Evaluate using RMSE, MAE, R².
• Save models to /models/.

Phase 5 — Zone-Level Aggregation

• Map Bengaluru into 223 geographic zones (GeoJSON).
• Aggregate predicted annual kWh and normalize into scores (0–100).
• Save as suitability_solar_geo.parquet.

Phase 6 — Visualization & Mapping

• Build GeoPandas + PyDeck layers for interactive visualization.
• Generate zone-level suitability map.

Phase 7 — Streamlit Dashboard

• Create final interactive app (app/streamlit_app.py):
  • Tabs: Suitability Map & Home Check
  • Top 10 Zones table
  • Full CSV download option
  • Home feasibility calculator with estimated annual energy

---

How to Run Locally

1. Clone this repository:

   git clone https://github.com/jaidevreddy/blr-renewables.git
   cd blr-renewables

2. pip install -r requirements.txt
3. streamlit run app/streamlit_app.py

---

Data Source

All climate and irradiance data used in this project comes from:
NASA POWER Project - thank you NASA