catalog.yml

metadata:
  version: 1
sources:
  # CA Open Data Portal has DCAT catalog
  ca_open_data:
    # We identify them as being loaded with the DCAT driver
    driver: dcat
    # Here we specify the args used to load the catalog
    args:
      # The URL to the catalog
      url: https://data.ca.gov/data.json
      # An optional name for the catalog
      name: ca_open_data
      # We can also specify a subset of the datasets in the catalog using an "items"
      # dictionary. If these are specified, only these datasets will be available in
      # the resulting catalog. They will be available under the more human-readable
      # name specified as the key.
      # Grab the identifier from the items we want.
      items:
        hospital_capacity: 529ac907-6ba1-4cb7-9aae-8966fc96aeef
  jhu_us_cases_parquet:
    driver: parquet
    description: US counties time-series data from Johns Hopkins University.
    args:
      urlpath: "../data/us-county-time-series.parquet"
  jhu_us_cases_csv:
    driver: csv
    description: US counties time-series data from Johns Hopkins University.
    args:
      urlpath: https://raw.githubusercontent.com/CityOfLosAngeles/covid19-indicators/master/data/us-county-time-series-short.csv
      csv_kwargs:
          dtype:
              fips: str
              date: str
  msa_county_crosswalk:
    driver: csv
    description: Crosswalk linking counties to metropolitan statistical areas.
    args:
      urlpath: https://raw.githubusercontent.com/CityOfLosAngeles/covid19-indicators/master/data/msa_county_pop_crosswalk.csv
      csv_kwargs:
        dtype:
            county_fips: str
            cbsacode: str
  # Other useful datasets
  jhu_global_cases:
    driver: csv
    description: Global country time-series data from Johns Hopkins University.
    args:
      urlpath: https://raw.githubusercontent.com/CityOfLosAngeles/covid19-indicators/master/data/global-time-series.csv
      csv_kwargs:
          dtype:
              number_of_recovered: float
  ca_hospital_surge_capacity:
    driver: csv
    description: CA county-level time-series data on hospitalizations and surge capacity.
    args:
      urlpath: https://raw.githubusercontent.com/CityOfLosAngeles/covid19-indicators/master/data/ca-hospital-and-surge-capacity.csv
  la_cases_by_neighborhood:
    driver: csv
    description: LA County neighborhood-level case data.
    args:
      urlpath: https://raw.githubusercontent.com/CityOfLosAngeles/covid19-indicators/master/data/la-county-neighborhood-time-series.csv
  # Crosswalks
  ca_counties_crosswalk:
    driver: csv
    description: Crosswalk linking CA counties with population.
    args:
      urlpath: https://raw.githubusercontent.com/CityOfLosAngeles/covid19-indicators/master/data/ca_county_pop_crosswalk.csv
  la_neighborhoods_crosswalk:
    driver: csv
    description: Crosswalk linking LA County's countywide statistical areas to larger neighborhoods with population.
    args:
      urlpath: https://raw.githubusercontent.com/CityOfLosAngeles/covid19-indicators/master/data/la_neighborhoods_population_crosswalk.csv