Dataset build scripts — PA 3311 / PS 3315
Dataset build scripts — PA 3311 / PS 3315
Reproducible code that generates the course datasets in ../data/.
| Script | Output | Description |
|---|---|---|
build_panel.py | TX_City_Sales_Panel_2013_2024.csv / .xlsx | Primary spine. Annual city-year panel (1,180 Texas cities × 2013–2024): sales-tax allocation, taxable sales, business outlets, sales-tax rate, plus time-invariant county / metro flag / 2022 reference population. |
build_dataset.py | TX_City_Finance_2022.csv / .xlsx | Cross-section supplement (one row per city, 2022) that adds debt and property/total taxes from the Census of Governments. |
See ../data/CODEBOOK.md for full variable definitions, sources, and the module-by-module usage map.
Data sources (all free, no API key)
- Texas Comptroller open data (
data.texas.gov): sales-tax allocation (vfba-b57j), quarterly taxable sales + outlets (7z4d-yf2c), city–county rate (53pa-m7sm). - U.S. Census Bureau, 2022 Census of Governments — Survey of Local Government Finances individual unit file (full enumeration; downloaded automatically to
_raw/).
Run
pip install pandas requests openpyxl
cd scripts
python3 build_panel.py # builds the panel
python3 build_dataset.py # builds the 2022 cross-section (with debt)
python3 validate.py # 19 integrity checks on both outputs (exit 0 = all pass)
Data provenance and licensing: see ../data/DATA_SOURCES.md. Each script downloads the Census file to scripts/_raw/ on first run and writes its CSV/XLSX to the current directory. The canonical copies committed to the repo live in ../data/; local rebuild outputs are git-ignored.