Spatial ML Benchmark

This project compares a majority baseline and a Random Forest using random and spatial evaluation. It includes reproducible runs and an offline map to inspect predictions and errors.

The demo uses a small extract of LUCAS Spain 2022: 32 points, two regions and elevation as the only predictor. You can switch evaluation methods and models, inspect metrics and select individual samples.

It is a learning and demonstration project. The sample does not support conclusions about national land-cover accuracy or statistical significance.