Deneb - Vega Lite Color Gradient Per Category

[DW868990]

Hi,

Just wondering if and how it is possible to set the color gradient for category within a visual rather than across the full visual.

The below example shows the gradient based all all categories but is is possible for the color gradient to be applied on each individual category.

{ "data": {"url": "data/cars.json"},

"mark": "rect",

"encoding": { "y": {"field": "Origin", "type": "nominal"},

"x": {"field": "Cylinders", "type": "ordinal"},

"color": {"aggregate": "mean", "field": "Horsepower"} },

"config": { "axis": {"grid": true, "tickBand": "extent"}

} }

Thanks in advance

[mgiammaria]

Hi @DW868990. Not quite sure if this is what you're after (see below), this implementation utilizes faceting. If it is, consider marking this as the solution. If not, could you please clarify what you mean? I'm making the assumption that you want your category to be Cylinders.

{
  "$schema": "https://vega.github.io/schema/vega-lite/v5.json",
  "data": {"url": "data/cars.json"},
  "transform": [
    {
      "aggregate": [{"op": "count", "as": "num_cars"}],
      "groupby": ["Origin", "Cylinders"]
    },
    {
      "window": [{"field": "num_cars", "op": "average", "as": "avg num_cars"}],
      "groupby": ["Cylinders"],
      "frame": [null, null]
    },
    {
      "window": [
        {"field": "num_cars", "op": "count", "as": "record ct per cylinders"}
      ],
      "groupby": ["Cylinders"],
      "frame": [null, null]
    },
    {
      "calculate": "((datum['num_cars']-datum['avg num_cars']))",
      "as": "variance"
    },
    {
      "sort": [
        {"field": "Cylinders"},
        {"field": "num_Cars", "order": "descending"}
      ]
    }
  ],
  "facet": {"column": {"field": "Cylinders", "title": null}},
  "title": {
    "text": "Count of Cars per Cylinder per Origin",
    "orient": "top",
    "anchor": "middle"
  },
  "spec": {
    "encoding": {
      "y": {"field": "Origin", "type": "ordinal"},
      "x": {"field": "Cylinders", "type": "ordinal", "axis": null}
    },
    "layer": [
      {
        "mark": "rect",
        "encoding": {
          "color": {
            "field": "num_cars",
            "type": "quantitative",
            "legend": {
              "direction": "vertical",
              "orient": "bottom",
              "gradientLength": 100,
              "title": null
            }
          }
        }
      },
      {
        "mark": "text",
        "encoding": {
          "text": {"field": "num_cars", "type": "quantitative"},
          "color": {
            "condition": {"test": "datum['variance'] < 0", "value": "black"},
            "value": "white"
          }
        }
      }
    ]
  },
  "resolve": {"scale": {"x": "independent", "color": "independent"}},
  "config": {"axis": {"grid": true, "tickBand": "extent"}}
}

[DW868990]

Thank you. Yes its nearly what i am after. Although the category i would be using is Origin in this example. 

So for example say Europe ranged from 1 to 100
and then USA ranged from 1 to 1000

The color for the highest value for each category would be the same gradient/colour.

Thanks for your help.

[giammariam]

@DW868990, how about this?

{
  "$schema": "https://vega.github.io/schema/vega-lite/v5.json",
  "data": {"url": "data/cars.json"},
  "transform": [
    {
      "aggregate": [{"op": "count", "as": "num_cars"}],
      "groupby": ["Origin", "Cylinders"]
    },
    {
      "window": [{"field": "num_cars", "op": "average", "as": "avg num_cars"}],
      "groupby": ["Origin"],
      "frame": [null, null]
    },
    {
      "window": [
        {"field": "num_cars", "op": "count", "as": "record ct per cylinders"}
      ],
      "groupby": ["Origin"],
      "frame": [null, null]
    },
    {
      "calculate": "((datum['num_cars']-datum['avg num_cars']))",
      "as": "variance"
    },
    {
      "sort": [
        {"field": "Cylinders"},
        {"field": "num_Cars", "order": "descending"}
      ]
    }
  ],
  "facet": {"row": {"field": "Origin", "title": null}},
  "title": {
    "text": "Count of Cars per Cylinder per Origin",
    "orient": "top",
    "anchor": "middle"
  },
  "spec": {
    "encoding": {
      "y": {"field": "Origin", "type": "ordinal", "axis": null},
      "x": {"field": "Cylinders", "type": "ordinal"}
    },
    "layer": [
      {
        "mark": "rect",
        "encoding": {
          "color": {
            "field": "num_cars",
            "type": "quantitative",
            "legend": {
              "direction": "horizontal",
              "orient": "right",
              "gradientLength": 100,
              "title": null
            }
          }
        }
      },
      {
        "mark": "text",
        "encoding": {
          "text": {"field": "num_cars", "type": "quantitative"},
          "color": {
            "condition": {"test": "datum['variance'] < 0", "value": "black"},
            "value": "white"
          }
        }
      }
    ]
  },
  "resolve": {"scale": {"y": "independent", "color": "independent"}},
  "config": {"axis": {"grid": true, "tickBand": "extent"}}
}

[DW868990]

@giammariam  Yes 99% there, aside from in my case its not an aggregation the gradients needs to be based on, its just a value of a measure. 

Thanks for your help.

[giammariam · Accepted solution]

@DW868990, oh wow, apologies. It was so similar to an example on the vega-lite page, that I missed what you were actually after. Let me know if this is better. Happy to make additional tweaks if necessary. I took out the text in the boxes to match what you have. Can add it back in though if you like.

{
  "$schema": "https://vega.github.io/schema/vega-lite/v5.json",
  "data": {"url": "data/cars.json"},
  "facet": {"row": {"field": "Origin", "title": null}},
  "title": {
    "text": "Mean Horsepower per Cylinder per Origin",
    "orient": "top",
    "anchor": "middle"
  },
  "spec": {
    "encoding": {
      "y": {"field": "Origin", "type": "ordinal", "axis": null},
      "x": {"field": "Cylinders", "type": "ordinal"}
    },
    "layer": [
      {
        "mark": "rect",
        "encoding": {
          "color": {
            "field": "Horsepower",
            "aggregate": "average",
            "type": "quantitative",
            "legend": {
              "direction": "horizontal",
              "orient": "right",
              "gradientLength": 100,
              "title": null
            }
          }
        }
      }
    ]
  },
  "resolve": {"scale": {"y": "independent", "color": "independent"}},
  "config": {"axis": {"grid": true, "tickBand": "extent"}}
}

[DW868990]

Fantastic, this is exactly what i was after, thank you.

[giammariam]

No problem. Thanks for sticking by through the iterations. Happy to help and get some Deneb practice in. Feel free to @ me in any deneb related questions you have in the future.