Aggregation

To aggregate data in Vega-Lite, users can either use the aggregate property of an encoding field definition or the aggregate transform inside the transform array. Aggregate summarizes a table as one record for each group. To preserve the original table structure and instead add a new column with the aggregate values, use the join aggregate transform.

Documentation Overview

• Aggregate in Encoding Field Definition
• Aggregate Transform
  • Aggregated Field Definition for Aggregate Transform
• Supported Aggregation Operations
• Argmin / Argmax
  • Example: Labeling Line Chart

Aggregate in Encoding Field Definition

// A Single View or a Layer Specification
{
  ...,
  "mark/layer": ...,
  "encoding": {
    "x": {
      "aggregate": ...,               // aggregate
      "field": ...,
      "type": "quantitative",
      ...
    },
    "y": ...,
    ...
  },
  ...
}

The aggregate property of a field definition can be used to compute aggregate summary statistics (e.g., median, min, max) over groups of data.

If at least one fields in the specified encoding channels contain aggregate, the resulting visualization will show aggregate data. In this case, all fields without aggregation function specified are treated as group-by fields¹ in the aggregation process.

For example, the following bar chart aggregates mean of Acceleration, grouped by the number of Cylinders.

Note: aggregated fields are quantitative by default while unaggregated (group by) fields in aggregated encodings are nominal by default.

The detail channel can be used to specify additional summary and group-by fields without mapping the field(s) to any visual properties. For example, the following plots add Origin as a group by field.

¹The group-by fields are also known as independent/condition variables in statistics and dimensions in Business Intelligence. Similarly, the aggregate fields are known as dependent variables and measures.

Aggregate Transform

// Any View Specification
{
  ...
  "transform": [
    {
      // Aggregate Transform
      "aggregate": [{"op": ..., "field": ..., "as": ...}],
      "groupby": [...]
    }
     ...
  ],
  ...
}

For example, here is the same bar chart which aggregates mean of Acceleration, grouped by the number of Cylinders, but this time using the aggregate property as part of the transform.

An aggregate transform in the transform array has the following properties:

Aggregated Field Definition for Aggregate Transform

Note: It is important you parse your data types explicitly, especially if you are likely to have null values in your dataset and automatic type inference will fail.

Supported Aggregation Operations

The supported aggregation operations are:

Argmin / Argmax

Sometimes, you may not want to find the minimum or maximum of a field, but instead the value from a field that corresponds to the minimum or maximum value in another field. In these cases you can use the argmin and argmax aggregates.

The argmax and argmin operation can be specified in an encoding field definition by setting aggregate to an object with argmax/min describing the field to maximize/minimize. For example, the following plot shows the production budget of the movie that has the highest US Gross in each major genre.

This is equivalent to specifying argmax in an aggregate transform and encode its nested data.

Example: Labeling Line Chart

argmax can be useful for getting the last value in a line for label placement.