censusdis

IMPORTANT NOTE [May 25, 2024]

The census bureau recently changed their server side code in a way that caused some issues for censusdis. We have updated censusdis to resolve the problem. Please be sure you are using version 1.1.8 or higher.

censusdis is a package for discovering, loading, analyzing, and computing diversity, integration, and segregation metrics to U.S. Census demographic data. It is designed

• to support every dataset, every geography, and every year. It's not just about ACS data through the last time the software was updated and released;
• to support all geographies, on and off-spine, not just states, counties, and census tracts;
• to have integrated mapping capabilities that save you time and extra coding;
• to be intuitive, Pythonic, and fast.

Click any of the thumbnails below to see the notebook that generated it.

Installation and First Example

censusdis can be installed with pip:

pip install censusdis

Every censusdis query needs four things:

1. What data set we want to query.
2. What vintage, or year.
3. What variables.
4. What geographies.

Here is an example of how we can use censusdis to download data once we know those four things.

import censusdis.data as ced
from censusdis.datasets import ACS5
from censusdis import states

df_median_income = ced.download(
    # Data set: American Community Survey 5-Year
    dataset=ACS5,
    
    # Vintage: 2022
    vintage=2022, 
    
    # Variable: median household income
    download_variables=['NAME', 'B19013_001E'], 
    
    # Geography: All counties in New Jersey.
    state=states.NJ,
    county='*'
)

There are many more examples in the tuturial and in the sample notebooks.

Tutorial (A Great Place to Start!)

We presented a half-day tutorial on censusdis at SciPy '24. All the material covered in the tutorial is available as in a github repo at https://github.com/censusdis/censusdis-tutorial-2024. The tutorial consists of a series of five lessons, each with worked exercises, and two choices for a final project. If you really want to learn the ins and outs of what censusdis can do, from the most basic queries all the way through some relatively advanced topics, this is the tutorial for you.

An Older Tutorial

For an older tutorial that is shorter but does not include some of the newest features, please see the censusdis-tutorial repository. This tutorial was presented at PyData Seattle 2023. If you want to try it out for yourself, the README.md contains links that let you run the tutorial notebooks live on mybinder.org in your browser without needing to set up a local development environment or download or install any code.

Tutorial Video

We expect a vireo of the SciPy '24 tutorial to be available soon, hopefully by some time in August '24.

A 86 minute video of the older tutorial as presented at PyData Seattle 2023 is also available.

Overview

censusdis is a package for discovering, loading, analyzing, and computing diversity, integration, and segregation metrics to U.S. Census demographic data. It is designed to be intuitive and Pythonic, but give users access to the full collection of data and maps the US Census publishes via their APIs. It also avoids hard-coding metadata about U.S. Census variables, such as their names, types, and hierarchies in groups. Instead, it queries this from the U.S. Census API. This allows it to operate over a large set of datasets and years, likely including many that don't exist as of time of this writing. It also integrates downloading and merging the geometry of geographic geometries to make plotting data and derived metrics simple and easy. Finally, it interacts with the divintseg package to compute diversity and integration metrics.

The design goal of censusdis are discussed in more detail in design-goals.md.

I'm not sure I get it. Show me what it can do.

The Nationwide Diversity and Integration notebook demonstrates how we can download, process, and plot a large amount of US Census demographic data quickly and easily to produce compelling results with just a few lines of code.

I'm sold! I want to dive right in!

To get straight to installing and trying out code hop over to our Getting Started guide.

censusdis lets you quickly and easily load US Census data and make plots like this one:

We downloaded the data behind this plot, including the geometry of all the block groups, with a single call:

import censusdis.data as ced
from censusdis.states import STATE_GA

# This is a census variable for median household income.
# See https://api.census.gov/data/2020/acs/acs5/variables/B19013_001E.html
MEDIAN_HOUSEHOLD_INCOME_VARIABLE = "B19013_001E"

gdf_bg = ced.download(
    "acs/acs5",  # The American Community Survey 5-Year Data
    2020,
    ["NAME", MEDIAN_HOUSEHOLD_INCOME_VARIABLE],
    state=STATE_GA,
    block_group="*",
    with_geometry=True
)

Similarly, we can download data and geographies, do a little analysis on our own using familiar Pandas data frame operations, and plot graphs like these

Modules

The public modules that make up the censusdis package are

Module	Description
censusdis.geography	Code for managing geography hierarchies in which census data is organized.
censusdis.data	Code for fetching data from the US Census API, including managing datasets, groups, and variable hierarchies.
censusdis.maps	Code for downloading map data from the US, caching it locally, and using it to render maps.
censusdis.states	Constants defining the US States. Used by the other modules.
censusdis.counties	Constants defining counties in all of the US States.

Demonstration Notebooks

There are several demonstration notebooks available to illustrate how censusdis can be used. They are found in the notebook directory of the source code.

The demo notebooks include

Notebook Name	Description
ACS Comparison Profile.ipynb	Load and plot American Community Survey (ACS) Comparison Profile data at the state level.
ACS Data Profile.ipynb	Load and plot American Community Survey (ACS) Data Profile data at the state level.
ACS Demo.ipynb	Load American Community Survey (ACS) Detail Table data for New Jersey and plot diversity statewide at the census block group level.
ACS Subject Table.ipynb	Load and plot American Community Survey (ACS) Subject Table data at the state level.
Block Groups in CBSAs.ipynb	Load and spatially join on-spine and off-spine geographies and plot the results on a map.
Congressional Districts.ipynb	Load congressional districts and tract-level data within them.
Data With Geometry.ipynb	Load American Community Survey (ACS) data for New Jersey and plot diversity statewide at the census block group level.
Exploring Variables.ipynb	Load metatdata on a group of variables, visualize the tree hierarchy of variables in the group, and load data from the leaves of the tree.
Geographies Contained within Geographies.ipynb	Demonstrate working with geograhies from different hierarchies.
Getting Started Examples.ipynb	Sample code from the Getting Started guide.
Nationwide Diversity and Integration.ipynb	Load nationwide demographic data, compute diversity and integration, and plot.
Map Demo.ipynb	Demonstrate loading at plotting maps of New Jersey at different geographic granularity.
Map Geographies.ipynb	Illustrates a large number of different map geogpraphies and how to load them.
Population Change 2020-2021.ipynb	Track the change in state population from 2020 to 2021 using ACS5 data.
PUMS Demo.ipynb	Load Public-Use Microdata Samples (PUMS) data for Massachusetts and plot it.
Querying Available Data Sets.ipynb	Query all available data sets. A starting point for moving beyond ACS.
Seeing White.ipynb	Load nationwide demographic data at the county level and plot of map of the US showing the percent of the population who identify as white only (no other race) at the county level.
SoMa DIS Demo.ipynb	Load race and ethnicity data for two towns in Essex County, NJ and compute diversity and integration metrics.
Time Series School District Poverty.ipynb	Demonstrates how to work with time series datasets, which are a little different than vintaged data sets.

Diversity and Integration Metrics

Diversity and integration metrics from the divintseg package are demonstrated in some notebooks.

For more information on these metrics see the divintseg project.