Documentation

Breaking Change (November 2022)

Due to free Dynos - which were used to proxy CORS requests - being deprecated by Heroku, pre 2.3 versions of CitySDK will cease to work client-side.

Additionally, the migration to AWS has forced us to migrate core config files which cause breaks in server-side code in the near future

Please update to the latest version of CitySDK (2.3) to fix

CitySDK v2

Thank You’s due to some very generous Clojurians:

• @thheller (author of the shadow-cljs build tool)
• @cgrand (author of the xforms library)
• The Clojure community at large

Installation

npm install citysdk

V 2.3 Changes

Starting with v2.3.0, CitySDK ships as an ESM export

Migration:

// 2.2.x or below
const census = require('citysdk')
// 2.3.x or above
import census from 'citysdk'

The citysdk Function

CitySDK exports a single function, which takes two arguments:

• The first is an options object with a set of key/value pair parameters (See “Parameters” below)
• The second is a conventional (error, response) node-style callback, which will be called upon completion of the census function and applied to the response

Parameters

Brief overview of each argument parameter that can be passed into CitySDK

* : optional ** : optional for < 500 requests daily

Geocoding (latitude/longitude -> FIPS code)

With the exception of “microdata” statistics (not yet available via Census’ API), all Census data is aggregated to geographic areas of different sizes. As such, all of Census’ API’s require a set of/unique geographic identifier(s) to return any data (AKA: FIPS). Given that these identifiers are not common knowledge, the CitySDK provides a way for the user to identify their geographic scope of interest using a geographic coordinate (lat + lng).

Under the hood, this functionality calls the TigerWeb Web Mapping Service with the lat & lng provided and pipes the resulting FIPS codes into your options argument with the appropriate GEOIDs for identifying your geographic area of interest.

For a list of geographies currently available for geocoding with this feature, see the Geographies Available by Vintage section below.

There are two ways to scope your geography using this functionality:

1. Request a single geographic area
2. Request all of a descendant geography-type of a coordinate-specified geographic area

Example: Request a single geographic area by coordinate

RETURN TYPE: JSON

You may pass a {"lat" : <float>, "lng" : <float>} object as the first and only value for the geoHierarchy key:

import census from 'citysdk'

census(
    {
        vintage: 2015, // required
        geoHierarchy: {
            // required
            county: {
                lat: 28.2639,
                lng: -80.7214,
            },
        },
    },
    (err, res) => console.log(res)
)

// result -> {"vintage":"2015","geoHierarchy":{"state":"12","county":"009"}}

Notice how the function prepends an additional geographic component ("state" : "12") to the options object. In order to fully qualify the geographic area (GEOID) associated with the county, the state is needed. In this example the fully qualified GEOID would be 12009 with the first two digits (12) qualifying the state and 009 qualifying the county within that state. This appropriate geographic hierarchy creation is handled by the function for you.

Example: Request all of a descendant geography-type within a coordinate-specified geographic area

RETURN TYPE: JSON

import census from 'citysdk'

census(
    {
        vintage: '2015', // required
        geoHierarchy: {
            // required
            state: {
                lat: 28.2639,
                lng: -80.7214,
            },
            county: '*', // <- syntax = "<descendant>" : "*"
        },
    },
    (err, res) => console.log(res)
)

// result -> {"vintage":"2015","geoHierarchy":{"state":"12","county":"*"}}

All Census-defined geographic areas are composed of Census “Blocks”. Some of these composed areas - themselves - compose into higher-order areas. These nested relationships between certain geographic areas allows the Census data user to request all descendants of a particular type.

👀 Caveats

1. Internally, the CitySDK converts the geoHierarchy object to an ordered set, so this part of your request object must be in descending hierarchical order from parent -> descendant. E.g. - in the above - an object that contained {"county" : "*", "state" : {"lat" <lat> "lng" <lng>}} will not work.
2. In this example, we added a second geographic level to our geoHierarchy object ("county" : "*"). It is important to use the "*" expression signifying that you want all of the specified level of descendants within the geography for which you supply a coordinate. No other expression will work.
3. For some wildcard ("*") geographies, the Census API can accept a skipped or “leapfrogged” wildcard. For example:

geoHierarchy: {
  state: "01",
  tract: "*"
}

However, the fully qualified geographic id requires an intermediary scope (in the above case county). You can tell when an intermediary scope has been skipped by checking the payload of the stats request who’s URL is logged by CitySDK.

Another indicator that you might be hitting this issue is if you get back an empty features list in your GeoJSON:

{ type: 'FeatureCollection', features: [ ] }

The solution to this problem is to add the skipped scope as a null property, e.g.:

geoHierarchy: {
  state: "01",
  county: null, // <- leapfrog fix
  tract: "*"
}

Statistics

This parameter set will call the Census Statistics API and reformat the results with a couple highly requested features:

• Census statistics are returned as a standard JSON object rather than the csv-like format of the “raw” API
• Statistical values are translated into properly typed numbers (Integers and Floats instead of strings), whereas all values are returned as strings via the “raw” API
• Annotation values (e.g., error codes) that are returned (e.g., American Community Survey error codes) in places where data would be expected are returned as strings (rather than numbers) to make differentiating them from values a simple type check.

There are two ways to request Census statistics using citysdk:

1. Calling for values of estimates and other statistical values (required)
2. Apply a filter by using predicates (optional)

For both of these options, a sourcePath needs to be supplied. This is the fully qualified path to the product. For more information about how to find the sourcePath to your product of interest, go to the Developers’ Microsite and - in any of the examples of making a call - take the path between <vintage>/ and the ?get. For example, for American Community Survey 1-year you’ll the first example (2017) shows:

https://api.census.gov/data/2017/acs/acs1?get=NAME,group(B01001)&for=us:1
                           └─┬─┘└───┬────┘
                         vintage sourcePath

The corresponding sourcePath for this endpoint is ["acs", "acs1"]

Example: get "values" by ID:

RETURN TYPE: JSON

census(
    {
        vintage: 2015, // required
        geoHierarchy: {
            // required
            county: {
                lat: 28.2639,
                lng: -80.7214,
            },
        },
        sourcePath: ['cbp'], // required
        values: ['ESTAB'], // required
    },
    (err, res) => console.log(res)
)

// result -> [{"ESTAB":13648,"state":"12","county":"009"}]

Here, we added the parameters for sourcePath (the path to the survey and/or source of the statistics) and values (the identifiers of the statistics we’re interested in). By including these parameters within your argument object, you trigger the census function to get statistics. This “deploy on parameter set” strategy is how the census function determines your intent.

🤔 Help for Discovering Census data

You’re probably thinking: “How am I supposed to know what codes to use inside those parameters?” - or - “Where did that "cbp" & "ESTAB" stuff come from?” The data sets covered by the CitySDK are vast. As such, this is the steepest part of the learning curve. But, don’t worry, there are a number of different resources available to assist you in your quest:

1. The Census Developers’ Microsite <- START HERE
2. The Census Discovery Tool.
3. Census Slack and Gitter developer communities.
4. Data Experts

Example: get "values" by ID (with key):

RETURN TYPE: JSON

census(
    {
        vintage: 2015, // required
        geoHierarchy: {
            // required
            county: {
                lat: 28.2639,
                lng: -80.7214,
            },
        },
        sourcePath: ['cbp'], // required
        values: ['ESTAB'], // required
        statsKey: '<your key here>', // required for > 500 calls per day
    },
    (err, res) => console.log(res)
)

// result -> [{"ESTAB":13648,"state":"12","county":"009"}]

Example: Filter results by predicates:

RETURN TYPE: JSON

predicates

Predicates are used to create a sub-selection of statistical values based on a given range or categorical qualifyer.

census(
    {
        vintage: '2017',
        geoHierarchy: {
            state: '51',
            county: '*',
        },
        sourcePath: ['acs', 'acs1'],
        values: ['NAME'],
        predicates: {
            B01001_001E: '0:100000', // number range separated by `:`
        },
        statsKey: '<your key here>',
    },
    (err, res) => console.log(res)
)

/* result:
    [
      {
        "NAME":"Augusta County, Virginia",
        "B01001_001E" : 75144,
        "state":"51",
        "county":"015"
      },
      {
        "NAME":"Bedford County, Virginia",
        "B01001_001E" : 77974,
        "state":"51",
        "county":"019"
      },
      ... 
    ]
*/

Timeseries data (Statistics Only)

If you’d like to use “timeseries” data, you may do so for statistics only. Mapping timeseries data is currently unsupported. Note that many timeseries products rely heavily on the "predicates" option:

Example: get 'timeseries" data:

RETURN TYPE: JSON

census(
    {
        vintage: 'timeseries', // required
        geoHierarchy: {
            // required
            us: '*',
        },
        sourcePath: ['asm', 'industry'], // required
        values: ['EMP', 'NAICS_TTL', 'GEO_TTL'],
        predicates: { time: '2016', NAICS: '31-33' },
    },
    (err, res) => console.log(res)
)

/* result:
[{"EMP": 11112764, 
  "NAICS_TTL": "Manufacturing", 
  "GEO_TTL": "United States", 
  "time": "2016", 
  "NAICS": "31-33", 
  "us":"1"}]
*/

For some sources (e.g., the American Community Survey), most of the values can also be used as predicates, but are optional. In others, (e.g., International Trade), predicates are a key part of the statistical query. In either case, at least one value within values must be supplied.

Cartographic GeoJSON

You can also use the CitySDK to retrieve Cartographic Boundary files, which have been translated into GeoJSON. The only additional parameter you’ll need to know is a simple declaration of geoResolution of which there are three options:

See the full available Cartographic GeoJSON in the Geographies Available by Vintage section

Example: Saving the file locally in Node.js using fs

RETURN TYPE: JSON STRING

import fs from "fs"

census(
    {
        vintage: 2017,
        geoHierarchy: {
            'metropolitan statistical area/micropolitan statistical area': '*',
        },
        geoResolution: '500k', // required
    },
    (err, res) => {
        fs.writeFile('./directory/filename.json', JSON.stringify(res), () => console.log('done'))
    }
)

This would convert the returned geojson to a string, which allows it to be saved via Node.js’ fileSystem API.

Notable Example:

census(
    {
        vintage: '2017',
        geoHierarchy: {
            state: '51',
            county: '*',
        },
        geoResolution: '500k', // required
    },
    (err, res) => console.log(res)
)

It’s important to note that - when querying for these GeoJSON files - you may retrieve a larger area than your request argument specifies. The reason for this is that the files are (currently) stored at two geographic levels: National and by State. Thus, the query above will attempt to resolve, at the state level, all counties, but because counties are stored at the national level in vintage 2017, all the counties in the US will be returned by this query.

If you wish to get back only those geographies you specify, you may do so by using the last and perhaps most useful feature included in the v2.0 release: Getting GeoJSON with statistics included within the "FeatureCollection" properties object!

GeoJSON Merged with Statistics

RETURN TYPE: JSON

There are a number of reasons you might want to merge your statistics into their GeoJSON/geographic boundaries, all of which are relevant when seeking to map Census data:

1. Creating choropleth maps of statistics (e.g., using values)
2. Mapping only those geographies that meet a certain set of criteria
3. Showing a user their current Census geographic context (i.e., leveraging the Geocoding capabilities of CitySDK)

Dynamic Use Example

A more dynamic example of using stats merged with GeoJSON on the fly with citysdk can be found here:

Type in a county name and see the unweighted sample count of the population (ACS) for all the Block Groups within that County.

Use Chrome for best results (mapbox-gl geocoder caveat)

source code

All Counties

census({
    vintage: '2017',
    geoHierarchy: {
        county: '*',
    },
    sourcePath: ['acs', 'acs5'],
    values: ['B19083_001E'], // GINI index
    statsKey: '<your key here>',
    geoResolution: '500k',
})

In this example, we use citysdk to create the payload and then save it via Nodes fs.writeFileSync and then serve it via a Mapbox-GL map.

source code

Notable Example:

All ZCTAs (zip code tabulation areas in the US)

census({
    vintage: '2017',
    geoHierarchy: {
        'zip-code-tabulation-area': '*',
    },
    sourcePath: ['acs', 'acs5'],
    values: ['B19083_001E'], // GINI index
    statsKey: '<your key here>',
    geoResolution: '500k',
})

This is a very large request, in fact, one of the largest you could possibly make in a single citysdk function call. It is so large, in fact that it currently only works on Node and only if you increase your node --max-old-space-size=4096. With large merges (such as all counties or zctas), it is recommended not to try to use citysdk dynamically, but - rather - to munge your data before hand with citysdk and then serve it statically to your mapping library, as was done here:

source code

Other Argument Examples:

// Call the WMS only
{
  "vintage": 2014,
  "geoHierarchy": { "state": { "lat": 28.2639, "lng": -80.7214 }, "county": '*' }
}

// Getting the stats for a single county filtering out any county with population under 100,000
{
  "vintage": 2016,
  "geoHierarchy": { "county": { "lat": 28.2639, "lng": -80.7214 } },
  "sourcePath": [ "acs", "acs5" ],
  "values": [ "B01001_001E" ]
  "predicates": { "B00001_001E": "0:100000" },
}

// strings are valid as vintages as well
{
  "vintage": "2015",
  "geoHierarchy": { "county": { "lat": 28.2639, "lng": -80.7214 } },
  "sourcePath": [ "cbp" ],
  "values": [ "ESTAB" ]
}

// Just geojson for all the counties within a state located by a given coordinate
{
  "vintage": 2014,
  "geoHierarchy": { "state": { "lat": 28.2639, "lng": -80.7214 }, "county": "*" },
  "geoResolution": "500k"
}

// For large request expect to have to increase `node --max-old-space-size=4096`
{
  "vintage": 2016,
  "sourcePath": [ "acs", "acs5" ],
  "values": [ "B25001_001E" ],
  "geoHierarchy": { "zip-code-tabulation-area": "*" },
  "geoResolution": "500k"
}

Census Cartography Files in GeoJSON Format

The Census Bureau publishes both high and low accuracy geographic area files to accommodate the widest possible variety of user needs (within feasibility). Cartography Files are simplified representations of selected geographic areas from the Census Bureau’s Master Address File/Topologically Integrated Geographic Encoding and Referencing (MAF/TIGER) system. These boundary files are specifically designed for small scale thematic mapping (i.e., for visualizations).

For a while now, we have published our cartography files in the .shp format. More recently, we expanded our portfolio of available formats to .kml. It is with this release that we follow suit with the community at large to release these boundaries in .json (GeoJSON) format.

Geographies Available by Vintage

The most comprehensive set of geographies and vintages can be found within the 500k set. Some vintages - 103 through 110 - are references to sessions of Congress and only contain a single geographic summary level: "congressional district" The following tables represent the availability of various geographic summary levels through the remaining vintages:

* = not available until Dec 2020

More Information about Cartography Files

• For more information about the files translated herein please visit the Census Bureau’s Cartographic Boundary File Description Page
• For a comparison of the available formats of geographic area files, please visit the Census Bureau’s TIGER Products page

Community

• Join us on Slack
• Send us an email: cnmp.developers.list@census.gov