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

This Zenodo record contains the reproducibility materials for the study “Frequent Mental Distress Across Texas Census Tracts: Social-Environmental Co-Exposure, Spatial Dependence, and Interpretable Machine Learning.” The package includes Jupyter notebooks for data-source auditing, land-cover processing, ACS socioeconomic data preparation, contextual variable construction, environmental exposure processing, analytic dataset assembly, ordinary least squares regression, spatial regression, sensitivity analyses, machine-learning validation, and SHAP interpretation. The primary analytic sample contains 6,802 Texas census tracts. A contextual extension contains 6,740 tracts after inclusion of mental-health-provider access. The workflow integrates CDC PLACES frequent mental distress estimates, ACS socioeconomic indicators, Annual NLCD 2023 imperviousness, EPA EJScreen air-pollution indicators, 2019–2023 extreme-heat exposure, USDA ERS RUCA classifications, County Health Rankings & Roadmaps provider-access data, and Texas Comptroller regional classifications. Processed analytic GeoPackages are included so that the statistical analyses can be reproduced without rebuilding all raw environmental source datasets. Large publicly available raw source files are not necessarily redistributed; their provenance and processing steps are documented in the included data-source table and notebooks.