Housing datasets

Average 30-year fixed-rate mortgage in the United States, weekly, April 1971-present.

S&P/Case-Shiller U.S. National Home Price Index (CSUSHPISA). Monthly data from 1987. Seasonally adjusted. THE standard home price index used in housing economics research, covering single-family home prices across the United States.

Consumer Price Index for rent of primary residence, not seasonally adjusted, monthly, 1914-present. Base period 1982-84=100. One of the longest-running FRED series. Separates rent from owner-equivalent rent in the broader shelter CPI.

Annual point-in-time counts of homeless persons by state from HUD's Annual Homeless Assessment Report (AHAR). Covers 2007-2025 with breakdowns by shelter type (emergency shelter, transitional housing, unsheltered), population type (individuals, families, family households), chronically homeless individuals, and homeless veterans. Based on a single-night count conducted in January each year by Continuums of Care (CoCs) nationwide. Source is a multi-sheet XLSB file where each year is a separate sheet with varying column counts (28 in 2007 to 1017 in 2025). Only the core summary columns consistent across all years are extracted via select_columns.

New privately-owned housing units started (HOUST). Monthly data from 1959. Seasonally adjusted annual rate in thousands of units. Key leading economic indicator for real estate and construction sectors.

Median sales price of new houses sold in the United States, quarterly, 1963-present. Nominal dollars (not inflation-adjusted). Covers new construction only, not existing home resales.

Typical home value for middle-tier single-family residences and condos, seasonally adjusted. Monthly data from January 2000 to present at state, metro, and county levels. Sourced from Zillow Research.

Household type breakdown for Texas counties from the Census Bureau's American Community Survey (ACS) 1-year estimates. Table B11001 covers family vs nonfamily households, married-couple vs other family types, and living arrangements. Years 2018-2023 (no 2020 data due to COVID-related ACS 1-year cancellation). Data stored in long format with Census variable codes; use the enriched view to see human-readable labels via the acs-variables dimension table.

Consumer Price Index for shelter (housing costs), seasonally adjusted, monthly, 1967-present. Base period 1982-84=100.

Annual 2-bedroom Fair Market Rents (FMR) by county/metro area from HUD, covering 1983 through 2026. FMRs set the payment standard for Housing Choice Vouchers (Section 8) and represent the 40th percentile of gross rents for standard-quality units. Covers ~4,764 counties and metro areas with FIPS codes for geographic joins. The 2-bedroom FMR is the standard HUD benchmark used for affordability analysis.

Licensed short-term rental (STR) properties in Austin, TX. Updated daily. Contains approximately 2,700 active STR licenses with property location, license type, and council district. Austin distinguishes between owner-occupied (Type 1), non-owner-occupied residential (Type 2), and non-owner-occupied multi-family (Type 3) short-term rentals. Addresses are shown at the block level for safety reasons.

Ames Iowa housing dataset from the Kaggle House Prices competition. Contains 1,460 homes with detailed features describing residential properties including size, quality, condition, and sale information. Popular regression benchmark.

California housing prices dataset from the 1990 census. Contains 20,640 districts with median house values and demographic features. One of the most popular regression benchmarks, used extensively in ML tutorials and scikit-learn documentation.

Construction permits issued for new construction and remodel projects in Austin since April 2016. A filtered subset of the full permits dataset with additional remodel valuation breakdowns. Approximately 477,000 records with permit type, work class, valuation details, square footage, and housing unit counts.

City of Austin affordable housing inventory with approximately 2,400 records. Includes project details, unit counts, income restriction levels (MFI tiers from 20% to 120%), affordability expiration dates, and funding sources (GO bonds, LIHTC, CDBG, HOME). Useful for tracking where affordable units exist, when restrictions expire, and which council districts have the most subsidized housing.

Boston housing prices dataset from Harrison and Rubinfeld (1978). Contains 506 census tracts with median home values and neighborhood characteristics. A classic regression benchmark, though deprecated in modern scikit-learn due to ethical concerns about the B variable.

DEPRECATED: This dataset is frozen at 2018 data from a static snapshot. Use fhfa/state-hpi for current state-level housing price index data. Federal Housing Finance Agency State House Price Index from 1975-2018. Monthly home price indices for all 50 US states compared to national average. Great for housing market analysis.

Code enforcement complaint cases from Austin Code Department, approximately 82,600 records. Includes case type, description, address, priority, inspector, and repeat offender and short-term rental flags. Cross-reference with STR locations and construction permit zones to explore whether development and short-term rentals drive complaints.

Residential demolition permits issued by the City of Austin. Approximately 11,500 records covering all residential teardowns with permit details, demolition category, council district, and location. High narrative value when cross-referenced with new construction permits to show the teardown-rebuild pipeline.

The headline Consumer Price Index and seven of its component indexes, stacked from eight FRED series into one long table with a `series` column. Each FRED dataset is a single monthly series with its own value column name; this table renames them all to `value` so one query can compare rent, shelter, food, medical care, transportation, tuition, and education against the all-items index. Series start in different years (rent back to 1914, education and communication only from 1993), so a given month has between one and eight rows. Every series is seasonally adjusted except rent, which FRED only publishes unadjusted.