Reproducibility materials

Manuscript: Persistence of cross-media toxic release profiles across U.S. counties
Public source: U.S. EPA Toxics Release Inventory Basic Plus files and reporting documentation

Contents
- project/08_tables: machine-readable values underlying the manuscript tables and reported diagnostics.
- project/numerical_summaries: compact numerical summaries produced by the analysis.
- code/build_analysis.py: deterministic base processing and analysis.
- code/build_extended_analysis.py: additional sensitivity and support analyses.

The public raw files are not redistributed in this compact archive. Obtain them from the agencies identified in the manuscript and place them under project/05_data/raw using the filenames documented by the source archive. From this archive root, run:

  python code/build_analysis.py
  python code/build_extended_analysis.py

Python dependencies are pandas, numpy, scipy, scikit-learn, matplotlib, and seaborn. The fixed random seed used for stochastic perturbation diagnostics is 20260816. Values published in the manuscript are also supplied directly in project/08_tables so they can be checked without downloading the raw archives.
