Data and code for: A Spatial Decision-Support Framework for Human-Bear Coexistence: Applying Machine Learning to Garbage-Driven Conflict in Florida

This repository contains the data and code supporting the paper “A Spatial Decision-Support Framework for Human-Bear Coexistence: Applying Machine Learning to Garbage-Driven Conflict in Florida.” We modeled garbage-related human–black bear conflict across Florida using Florida Fish and Wildlife Conservation Commission (FWC) bear reports (2014–2023) and socio-ecological predictors, with MaxEnt, Random Forest, CatBoost, and an AUC-weighted ensemble. The ensemble risk was combined with land-use-based intervention potential to create the Conflict Risk-Action Priority Index (CRA-PI), and its robustness was tested with a Monte Carlo sensitivity analysis. The repository includes the occurrence data, predictor values at presence and pseudo-absence points, and the R and Python scripts for spatial thinning, modeling, ensemble prediction, evaluation, and sensitivity analysis. Each folder contains a README describing its files. Large raster layers are not included because of file size and are available from the corresponding author on request.