datasets.yml

datasets:

  - name: citibike
    title: NYC Citibike trips 2018
    description:
    lastUpdated:

    backend:
      counts:
        table: nyc_citibike_trips
        columns:
          origin: start_station_id
          dest: end_station_id
          count: 1
      locations:
        type: csv
        file: citibike-stations.csv
        columns:
          id: id
          name: name
          latitude: lat
          longitude: lon

    attributes:

      - name: hour
        type: numeric
        label: Hour
        expression: toHour(start_time)

      - name: month
        type: category
        label: Month
        expression: toString(toMonth(start_date))
        values:
          - value: "1"
            label: Jan
          - value: "2"
            label: Feb
          - value: "3"
            label: Mar
          - value: "4"
            label: Apr
          - value: "5"
            label: May
          - value: "6"
            label: Jun
          - value: "7"
            label: Jul
          - value: "8"
            label: Aug
          - value: "9"
            label: Sep
          - value: "10"
            label: Oct
          - value: "11"
            label: Nov
          - value: "12"
            label: Dec

      - name: weekday
        type: category
        label: Weekday
        expression: toString(toDayOfWeek(start_date))
        values:
          - value: "1"
            label: Mon
          - value: "2"
            label: Tue
          - value: "3"
            label: Wed
          - value: "4"
            label: Thu
          - value: "5"
            label: Fri
          - value: "6"
            label: Sat
          - value: "7"
            label: Sun

      - name: trip_duration
        label: Trip Duration
        type: numeric

      - name: user_type
        label: User Type

      - name: gender
        label: Gender
        values:
          - value: "1"
            label: Male
          - value: "2"
            label: Female
          - value: "0"
            label: Other

      - name: birth_year
        label: Birth Year
        type: numeric
        expression: ROUND(birth_year/10)*10