Sorting rows of a faceted chart in vega lite - vega-lite

Question

Is there some magic to sort sub-charts from a row encoding by the number of items per row?
In this example with the cars dataset, i would want to have USA at the top, because it contains the most items, then Japan, then Europe:
Maybe with the items property of the row header data store?

Answer 1

I used aggregate transform to calculate the number of items per Origin and Cylinder and a subsequent join aggregate transformation to sum up the number of items per Origin:
"transform": [{
"aggregate": [{"op": "count", "as": "Count"}],
"groupby": ["Cylinders", "Origin"]
},
{
"joinaggregate": [{"op": "sum", "field": "Count", "as": "OriginCount"}],
"groupby": ["Origin"]
}
]
I can then display Count on the x-axis as any other data variable and sort the row encoding by the calculated OriginCount:
"encoding": {
"row": {
"field": "Origin", "type": "nominal",
"sort": { "field": "OriginCount", "order": "descending"}
},
"x": {
"field": "Count", "type": "quantitative"
},
...
]
Giving me the following grouped bar chart:
See in the Vega Editor
The full spec for reference:
{
"$schema": "https://vega.github.io/schema/vega-lite/v4.json",
"data": { "url": "https://vega.github.io/editor/data/cars.json"},
"transform": [{
"aggregate": [{"op": "count", "as": "Count"}],
"groupby": ["Cylinders", "Origin"]
},
{
"joinaggregate": [{"op": "sum", "field": "Count", "as": "OriginCount"}],
"groupby": ["Origin"]
}
],
"mark": {
"type": "bar",
"tooltip": true
},
"width": 400,
"encoding": {
"row": {
"field": "Origin", "type": "nominal",
"sort": { "field": "OriginCount", "order": "descending"}
},
"x": {
"field": "Count", "type": "quantitative"
},
"y": {
"field": "Cylinders", "type": "nominal"
}
}
}