Context

This database contains the data reported in the Annual Homeless Assessment Report to Congress (AHAR). It represents a point-In-time count (PIT) of homeless individuals, as well as a housing inventory count (HIC) conducted annually.

The data represent the most comprehensive national-level assessment of homelessness in America, including PIT and HIC estimates of homelessness, as well as estimates of chronically homeless persons, homeless veterans, and homeless children and youth.

These data can be trended over time and correlated with other metrics of housing availability and affordability, in order to better understand the particular type of housing resources that may be needed from a social determinants of health perspective.

HUD captures these data annually through the Continuum of Care (CoC) program. CoC-level reporting data have been crosswalked to county levels for purposes of analysis of this dataset.

Querying BigQuery tables

You can use the BigQuery Python client library to query tables in this dataset in Kernels. Note that methods available in Kernels are limited to querying data. Tables are at bigquery-public-data.sdoh_hud_pit_homelessness

Sample Query

What has been the change in the number of homeless veterans in the state of New York’s CoC Regions since 2012? Determine how the patterns of homeless veterans have changes across the state of New York

homeless_2018 AS ( SELECT Homeless_Veterans AS Vet18, CoC_Name
FROM bigquery-public-data.sdoh_hud_pit_homelessness.hud_pit_by_coc WHERE SUBSTR(CoC_Number,0,2) = "NY" AND Count_Year = 2018 ),

veterans_change AS ( SELECT homeless_2012.COC_Name, Vet12, Vet18, Vet18 - Vet12 AS VetChange FROM homeless_2018 JOIN homeless_2012 ON homeless_2018.CoC_Name = homeless_2012.CoC_Name )

SELECT * FROM veterans_change

The graph displays the top 15 states by an estimated number of homeless people in the United States for the year 2025. The x-axis represents U.S. states, while the y-axis shows the number of homeless individuals in each state. California has the highest homeless population with 187,084 individuals, followed by New York with 158,019, while Hawaii places last in this dataset with 11,637. This bar graph highlights significant differences across states, with some states like California and New York showing notably higher counts compared to others, indicating regional disparities in homelessness levels across the country.

This dataset contains estimates of homelessness, as well as estimates of chronically homeless persons, homeless veterans, and homeless children and youth provided by The U.S. Department of Housing and Urban Development. The estimates cover the period of years 2007-2017 and are at national, state and Continuums of Care (CoC) Point-In-Time (PIT) level.

This report outlines the key findings of the annual Point-In-Time (PIT) count and Housing Inventory Count (HIC) conducted in January of each year. Specifically, this report provides estimates of homelessness self-reported, as well as estimates of chronically homeless persons, homeless veterans, and homeless children and youth.

Current link at the time of dataset creation: https://www.hudexchange.info/resource/4832/2015-ahar-part-1-pit-estimates-of-homelessness/

Comprehensive dataset containing 3,632 verified Homeless shelter businesses in United States with complete contact information, ratings, reviews, and location data.

Abstract copyright UK Data Service and data collection copyright owner. A comparative study of the causes of new episodes of homelessness among people aged 50 or more years was undertaken in Boston, Massachusetts (USA), Melbourne, Australia, and four English cities. The aims were to make a substantial contribution to the predominantly American debate on the causes of homelessness, and to make practice recommendations for the improvement of prevention. The study had several objectives. It aimed to collect information about the antecedents, triggers and risk factors for becoming homeless in later life and about the national and local policy and service contexts. Furthermore, the researchers aimed to analyse and interpret the findings with reference to an integrated model of the causes of homelessness that represented structural and policy factors, including housing, health and social service organisation and delivery factors, and personal circumstances, events, problems and dysfunctions. The aim was to do this collaboratively, by drawing on the project partners' experience and knowledge. Finally, it was hoped to develop recommendations for housing, primary health care and social welfare organisations for the prevention of homelessness. This was to be done by identifying the common sequences and interactions of events that precede homelessness and their markers (or 'early warning' indicators) and by holding workshops in England with practitioners and their representative organisations on new ways of working. By the study of contrasting welfare and philanthropic regimes in a relatively homogeneous category of homeless incidence (i.e. recent cases among late middle-aged and older people), it was hoped that valuable insights into the relative contributions of the policy, service and personal factors would be obtained. The study focused on older people who had recently become homeless, purposely to gather detailed and reliable information about the prior and contextual circumstances. To have included people who had been homeless for several years would have reduced the quality of the data because of 'recall' problems. Users should note that data from the Australian sample for the study are not included in this dataset. Main Topics: The data file includes information about the English respondents and those from Boston. It was compiled in two stages. The first stage involved each project partner entering the pre-coded responses into the file. All partners then identified themes and created codes for the open-ended responses, and the resulting variables were added. Data quality-control procedures included blind checks of the data coding and keying. The first 200 variables pertain to information collected from the respondents. They comprise descriptive variables of the circumstances prior to homelessness, including housing tenure during the three years prior to the survey, previous homelessness, employment history, income, health and addiction problems, and contacts with family, friends and formal services. The respondents were asked to rate whether specific factors were implicated in becoming homeless, and where appropriate, a following open-ended question sought elaboration. The remaining variables comprise information collected from the respondents' 'key workers' about their understanding of the events and states that led to their clients becoming homeless. No sampling frame was available. The sample profiles have been compared with those of all homeless people (not just the recently homeless) in the study locations, most effectively in London and Boston. No gross biases were revealed. The samples represent a large percentage of the clients who presented to the collaborating organisations during the study period and who gave their informed consent to participate. Agreed definitions of homelessness were: sleeping on the streets or in temporary accommodation such as shelters; being without accommodation following eviction or discharge from prison or hospital; living temporarily with relatives or friends because the person has no accommodation, but only if the stay had not exceeded six months, and the person did not pay rent and was required to leave. People who had been previously homeless were included in the survey if they had been housed for at least 12 months prior to the current episode of homelessness. Face-to-face interview Self-completion the 'key workers' (case managers) completed questionnaires about their assessments of the respondents’ problems and of the events and states that led to homelessness. Further clarifications and checks were made by telephone.

Compares demographic characteristics of COVID respite and hotel shelters operating March 2020 to June 2022 with all single adults in shelters and total homeless population from the point-in-time count on 1/23/2020 and total county population from the US Census American Community Survey for 2018. Multiracial American Indians are counted in "American Indian" for the respite and hotel shelters and singles shelters whereas all multiracial persons in other data sets are counted in the category "Multiracial".

The index is constructed using socioeconomic and demographic, exposure, health, and housing indicators and is intended to serve as a planning tool for health and climate adaptation. Steps for calculating the index can be found in in the "An Assessment of San Francisco’s Vulnerability to Flooding & Extreme Storms" located at https://sfclimatehealth.org/wp-content/uploads/2018/12/FloodVulnerabilityReport_v5.pdf.pdfData Dictionary: (see attachment here also: https://data.sfgov.org/Health-and-Social-Services/San-Francisco-Flood-Health-Vulnerability/cne3-h93g)

Field Name Data Type Definition Notes (optional)

Census Blockgroup Text San Francisco Census Block Groups

Children Numeric Percentage of residents under 18 years old. American Community Survey 2009 - 2014.

Chidlren_wNULLvalues Numeric Percentage of residents under 18 years old. American Community Survey 2009 - 2014. Because the American Community Survey uses survey estimates, all data is attached to a margin of error. When the coefficient of variation is over .3, the SFDPH considers this data unstable and gives it a NULL value. However, because principal component analysis and the final development of the flood health index could not use NULL values, SFDPH used this unstable data for these limited purposes. For the purpose of transparency, SFDPH has included both datasets with NULL values and without NULL values.

Elderly Numeric Percentage of residents aged 65 and older. American Community Survey 2009 - 2014.

Elderly_wNULLvalues Numeric Percentage of residents aged 65 and older. American Community Survey 2009 - 2014. Because the American Community Survey uses survey estimates, all data is attached to a margin of error. When the coefficient of variation is over .3, the SFDPH considers this data unstable and gives it a NULL value. However, because principal component analysis and the final development of the flood health index could not use NULL values, SFDPH used this unstable data for these limited purposes. For the purpose of transparency, SFDPH has included both datasets with NULL values and without NULL values.

NonWhite Numeric Percentage of residents that do not identify as white (not Hispanic or Latino). American Community Survey 2009 - 2014.

NonWhite_wNULLvalues Numeric Percentage of residents that do not identify as white (not Hispanic or Latino). American Community Survey 2009 - 2014. Because the American Community Survey uses survey estimates, all data is attached to a margin of error. When the coefficient of variation is over .3, the SFDPH considers this data unstable and gives it a NULL value. However, because principal component analysis and the final development of the flood health index could not use NULL values, SFDPH used this unstable data for these limited purposes. For the purpose of transparency, SFDPH has included both datasets with NULL values and without NULL values.

Poverty Numeric Percentage of all individuals below 200% of the poverty level. American Community Survey 2009 - 2014.

Poverty_wNULLvalues Numeric Percentage of all individuals below 200% of the poverty level. American Community Survey 2009 - 2014. Because the American Community Survey uses survey estimates, all data is attached to a margin of error. When the coefficient of variation is over .3, the SFDPH considers this data unstable and gives it a NULL value. However, because principal component analysis and the final development of the flood health index could not use NULL values, SFDPH used this unstable data for these limited purposes. For the purpose of transparency, SFDPH has included both datasets with NULL values and without NULL values.

Education Numeric Percent of individuals over 25 with at least a high school degree. American Community Survey 2009 - 2014.

Education_wNULLvalues Numeric Percent of individuals over 25 with at least a high school degree. American Community Survey 2009 - 2014. Because the American Community Survey uses survey estimates, all data is attached to a margin of error. When the coefficient of variation is over .3, the SFDPH considers this data unstable and gives it a NULL value. However, because principal component analysis and the final development of the flood health index could not use NULL values, SFDPH used this unstable data for these limited purposes. For the purpose of transparency, SFDPH has included both datasets with NULL values and without NULL values.

English Numeric Percentage of households with no one age 14 and over who speaks English only or speaks English "very well". American Community Survey 2009 - 2014.

English_wNULLvalues Numeric Percentage of households with no one age 14 and over who speaks English only or speaks English "very well". American Community Survey 2009 - 2014. Because the American Community Survey uses survey estimates, all data is attached to a margin of error. When the coefficient of variation is over .3, the SFDPH considers this data unstable and gives it a NULL value. However, because principal component analysis and the final development of the flood health index could not use NULL values, SFDPH used this unstable data for these limited purposes. For the purpose of transparency, SFDPH has included both datasets with NULL values and without NULL values.

Elevation Numeric Minimum elevation in feet. United States Geologic Survey 2011.

SeaLevelRise Numeric Percent of land area in the 100-year flood plain with 36-inches of sea level rise. San Francisco Sea Level Rise Committee, AECOM 77inch flood inundation layer, 2014.

Precipitation Numeric Percent of land area with over 6-inches of projected precipitation-related flood inundation during an 100-year storm. San Francisco Public Utilities Commission, AECOM, 2015.

Diabetes Numeric Age-adjusted hospitalization rate due to diabetes; adults 18+. California Office of Statewide Health Planning and Development, 2004-2015.

MentalHealth Numeric Age-adjusted hospitalization rate due to schizophrenia and other psychotic disorders. California Office of Statewide Health Planning and Development, 2004-2015.

Asthma Numeric Age-adjusted hospitalization rate due to asthma; adults 18+. California Office of Statewide Health Planning and Development, 2004 - 2015.

Disability Numeric Percentage of total civilian noninstitutionalized population with a disability. American Community Survey 2009 - 2014.

Disability_wNULLvalues

Percentage of total civilian noninstitutionalized population with a disability. American Community Survey 2009 - 2014. Because the American Community Survey uses survey estimates, all data is attached to a margin of error. When the coefficient of variation is over .3, the SFDPH considers this data unstable and gives it a NULL value. However, because principal component analysis and the final development of the flood health index could not use NULL values, SFDPH used this unstable data for these limited purposes. For the purpose of transparency, SFDPH has included both datasets with NULL values and without NULL values.

HousingQuality Numeric Annual housing violations, per 1000 residents. San Francisco Department of Public Health, San Francisco Department of Building Inspections, San Francisco Fire Department, 2010 - 2012.

Homeless Numeric Homeless population, per 1000 residents. San Francisco Homeless Count 2015.

LivAlone Numeric Households with a householder living alone. American Community Surevey 2009 - 2014.

LivAlone_wNULLvalues Numeric Households with a householder living alone. American Community Surevey 2009 - 2014. Because the American Community Survey uses survey estimates, all data is attached to a margin of error. When the coefficient of variation is over .3, the SFDPH considers this data unstable and gives it a NULL value. However, because principal component analysis and the final development of the flood health index could not use NULL values, SFDPH used this unstable data for these limited purposes. For the purpose of transparency, SFDPH has included both datasets with NULL values and without NULL values.

FloodHealthIndex Numeric Comparative ranking of flood health vulnerability, by block group. The Flood Health Index weights the six socioeconomic and demographic indicators (Children, Elderly, NonWhite, Poverty, Education, English) as 20% of the final score, the three exposure indicators (Sea Level Rise, Precipitation, Elevation) as 40% of the final score, the four health indicators (Diabetes, MentalHealth, Asthma, Disability) as 20% of the final score, and the three housing indicators (HousingQuality, Homeless, LivAlone) as 20% of the final score. For methodology used to develop the final Flood Health Index, please read the San Francisco Flood Vulnerability Assessment Methodology Section.

FloodHealthIndex_Quintiles Numeric Comparative ranking of flood health vulnerability, by block group. The Flood Health Index weights the six socioeconomic and demographic indicators (Children, Elderly, NonWhite, Poverty, Education, English) as 20% of the final score, the three exposure indicators (Sea Level Rise, Precipitation, Elevation) as 40% of the final score, the four health indicators (Diabetes, MentalHealth, Asthma, Disability) as 20% of the final score, and the three housing indicators (HousingQuality, Homeless, LivAlone) as 20% of the final score. For methodology used to develop the final Flood Health Index, please read the San Francisco Flood