ML-Based Climate Hotspot Analysis: Code and Data for Sindh Province, Pakistan

This repository contains the complete code and data for a machine learning-based climate hotspot analysis of Sindh Province, Pakistan. The study integrates multiple climate datasets (NASA POWER, EM-DAT, IBTrACS) with human activity indicators (nightlights, NDVI, emissions, population) to identify climate-vulnerable districts using ensemble ML methods (XGBoost, LightGBM) with SHAP-based explainability. The analysis includes IPCC risk assessment, Paris Agreement compliance evaluation, and Mann-Kendall trend testing across 22 administrative districts. This dataset accompanies a manuscript submitted to Geoscientific Model Development (GMD).