utah-housing 0.1.0

Census ACS data pull and fixed effects models for Utah housing research

utah-housing

Census ACS 5-year data fetcher and fixed effects models for Utah housing research.

Setup

pip install -e ".[dev]"
export CENSUS_API_KEY=your_key_here  # https://api.census.gov/data/key_signup.html

Usage

Pull data

from utah_housing import fetch_all_years

df = fetch_all_years(years=range(2009, 2024))
df.to_csv("utah_housing_2009_2023.csv", index=False)

Pull a single year:

from utah_housing import fetch_year

df_2022 = fetch_year(2022)

Run models

import pandas as pd
from utah_housing import run_base_model, run_complex_model, compare_models

df = pd.read_csv("utah_housing_2009_2023.csv")

result1, coef1 = run_base_model(df)
result2, coef2 = run_complex_model(df)

comparison = compare_models(coef1, coef2)
coef1.to_csv("base_results.csv")
coef2.to_csv("complex_results.csv")

Run diagnostics

from utah_housing.models import run_diagnostics
from utah_housing import COMPLEX_PREDICTORS

run_diagnostics(df, COMPLEX_PREDICTORS)

Models

Model	Fixed effects	Extra predictor
Base	tract + year	—
Complex	tract + county×year	pop_in_occupied_total

Outcome: median_owner_costs_with_mortgage

Predictors:

• pct_sf_renter_occupied — share of single-family homes that are renter-occupied (investor proxy)
• median_household_income — demand-side income
• owner_renter_income_gap — income stratification signal
• pct_vacant — market slack

Data tables pulled

Table	Description
B25024	Units in structure
B25001	Total housing units
B25002	Occupancy status
B25003	Tenure (owner vs. renter)
B25008	Population in occupied housing by tenure
B25119	Median household income by tenure
B25032	Units in structure by tenure
B25088	Median monthly owner costs
B19013	Median household income

Package layout

utah_housing/
├── __init__.py      # public API
├── variables.py     # all ACS variable lists, rename map, model variable sets
├── fetch.py         # Census API fetcher
└── models.py        # fixed effects models + diagnostics