census-shapefile-utils

Tools for fetching shapefiles from the Census FTP site, then extracting data from them.

Installation

1. Clone this repository: [email protected]:censusreporter/census-shapefile-utils.git
2. Enter the census-shapefile-utils directory.
3. If using Python 3 or parse_shapefiles.py, then install dependencies: pip install -r requirements.txt
   • Note: package gdal requires non-Python library, libgdal. Follow OS-specific installation to obtain this library.

fetch_shapefiles.py

This script will download TIGER data shapefiles from the Census FTP site. It can be used to download a set of geographies defined in GEO_TYPES_LIST, or can be used to fetch files for a single state and/or single geography type. Pass an -s argument to limit by state, pass a -g argument to limit to a single geography type, and/or pass a -y argument to change the year from 2012 to something else (e.g. 2015).

>> python fetch_shapefiles.py
>> python fetch_shapefiles.py -s WA
>> python fetch_shapefiles.py -g place
>> python fetch_shapefiles.py -y 2015
>> python fetch_shapefiles.py -s WA -g place -y 2015

If you use the -s argument to fetch files for a single state, the script will also download the national county, state and congressional district files that include data for your chosen state.

The script will create DOWNLOAD_DIR and EXTRACT_DIR directories if necessary, fetch a zipfile or set of zipfiles from the Census website, then extract the shapefiles from each zipfile retrieved.

DISABLE_AUTO_DOWNLOADS will prevent certain geography types from being automatically downloaded if no -g argument is passed to fetch_shapefiles.py. This may be useful because certain files, such as those for Zip Code Tabulation Areas, are extremely large. You can still target any geography in GEO_TYPES_LIST specifically, however. So to fetch the ZCTA data:

>> python fetch_shapefiles.py -g zcta5

parse_shapefiles.py

After you run fetch_shapefiles.py, this script will generate a csv file from the extracted data. These files will have a normalized set of headers, so varying geography types can be included in the same csv. Each row also gets some useful additional fields not directly found in the shapefiles, Including:

• FULL_GEOID: Concatenated Census summary level code and Census GEOID
• FULL_NAME: Human-friendly name for the geography. So city names, for instance, also include the state name, e.g. "Spokane, Washington"
• SUMLEV: Census summary level code
• GEO_TYPE: Name of the geography type, e.g. "state"
• REGION: Where applicable, a Census Region code. Shapefiles for states include this code; this script infers the value based on state for other geography types.
• REGION_NAME: Name of the Census Region, e.g. "West"
• DIVISION: Where applicable, a Census Division code. Shapefiles for states include this code; this script infers the value based on state for other geography types.
• DIVISION_NAME: Name of the Census Division, e.g. "Pacific"

This script will search all directories inside EXTRACT_DIR for shapefiles. Pass an -s argument to limit by state, and/or pass a -g argument to limit to a single geography type.

>> python parse_shapefiles.py
>> python parse_shapefiles.py -s WA
>> python parse_shapefiles.py -g place
>> python parse_shapefiles.py -s WA -g place

This script will generate a single csv file with your chosen data, and write it to CSV_DIR. Headers are pulled from helpers/csv_helpers.py. The methods for building rows specific to each geography type are also in csv_helpers.

You can choose whether the generated csv should include polygon geometries, which can significantly increase the size of the output file. Include geometries by passing a -p flag.

>> python parse_shapefiles.py -s WA -p

Geometry data for certain geography types can be very large. The zcta5 geometries, for instance, will add about 1.1 Gb of data to your csv.

Examples

These assume you have already used fetch_shapefiles.py to download the shapefiles you want to get data from.

>> python parse_shapefiles.py -g place will make place.csv, which includes data from all records at the Census place level.

>> python parse_shapefiles.py -s WA will make all_geographies_WA.csv, which includes all geographies in Washington state, from the state record all the way down to cities (places) and school districts. It will not include polygon geometries.

>> python parse_shapefiles.py -s WA -g county -p will make county_WA.csv, which includes data from all counties in Washington state. Because the -p flag was passed, it will also include polygon geometries for each record.

>> python parse_shapefiles.py will make all_geographies.csv, which includes data from all geography levels and all states. If you've downloaded shapefiles for all levels, including for Zip Code Tabulation Areas, the csv file should be about 19 Mb.

>> python parse_shapefiles.py -p will make the same file as above, but including geometries. This file takes about 17 minutes to build locally on my Macbook Air, and is about 2.45 Gb.