📌 Overview

This project implements an end-to-end geospatial machine learning pipeline for classifying tree species using Sentinel-1 SAR and Sentinel-2 optical data.

🚀 Features

• Feature extraction from Sentinel-1 and Sentinel-2.
• Vegetation indices (NDVI, EVI, SAVI).
• Integration of terrain and climate data.
• ML models: Random Forest, XGBoost, LightGBM.

⚙️ Workflow

1. Extract features using Google Earth Engine.
2. Perform preprocessing and feature engineering.
3. Train ML models and evaluate performance.
4. Deploy best-performing model.

📊 Results

Best Model: XGBoost
Accuracy: 75%

🛠 Tools & Technologies

Python, Google Earth Engine, Remote Sensing, scikit-learn

📌 Overview

This project develops a data-driven decision support system that recommends optimal tree species for agroforestry systems using environmental and spatial data. It integrates machine learning and remote sensing to support climate-smart agriculture and sustainable land use planning.

🚀 What Makes It Unique

• Combines remote sensing, climate, and terrain data into one ML pipeline.
• Transforms research into a deployed real-world application.
• Provides location-specific recommendations for farmers and policymakers.

⚙️ How It Works

1. Extract environmental variables (climate, elevation, vegetation indices).
2. Train ML models (RF, SVM, GBM, MaxEnt).
3. Generate habitat suitability predictions.
4. Deploy via an interactive Shiny web application.

📊 Key Results

• Random Forest achieved AUC > 0.99.
• High recall and specificity for ecological predictions.
• Successfully deployed as a real-world decision tool.

🛠 Tech Stack

Google Earth Engine, Python, R (Shiny, Plumber API), Machine Learning, Remote Sensing

• 🚀 Launch Live App