<p dir="ltr">This repository tackles **complexity metric computation, classification modeling, and feature ablation for raster data**. The workflow covers:</p><p dir="ltr">1. Batch-reading `.tif` rasters, computing 19 spatial complexity metrics, and exporting the results;</p><p dir="ltr">2. Training Random Forest, XGBoost, and LightGBM classifiers on the metrics with multiple feature-selection strategies;</p><p dir="ltr">3. Running SHAP-based ablation studies to measure how feature combinations influence accuracy and efficiency.</p>
