Data and code for: Frequent Mental Distress Across Texas Census Tracts: Social-Environmental Co-Exposure, Spatial Dependence, and Interpretable Machine Learning

This repository contains the processed analytic dataset and analysis code supporting the study "Environmental Co-Exposure, Green Space, and Frequent Mental Distress in Texas Census Tracts: An Interpretable Machine Learning and Spatial Analysis." The dataset includes tract-level frequent mental distress estimates (CDC PLACES) linked with socioeconomic variables (American Community Survey 2019–2023), land-cover indicators (National Land Cover Database via IPUMS NHGIS), air-pollution variables (CDC PM2.5 and EPA EJScreen), and heat exposure metrics (PRISM daily maximum temperature) for 6,802 Texas census tracts. The accompanying Jupyter notebook contains the full analysis workflow, including descriptive statistics, staged ordinary least squares regression, spatial autocorrelation diagnostics, spatial regression models, Random Forest modeling with SHAP interpretation, and county-grouped cross-validation.