data.group

data.group partitions rows by one or more fields, then computes aggregation functions over each group. The output contains one row per unique combination of group-by values, plus the aggregation result columns.

Input: Table Output: Table (group columns + aggregation columns)

Basic config

nodes:
  orders-by-status:
    type: data.group
    config:
      by: [status]
      aggregations:
        order_count:
          op: count
          field: id
        total_revenue:
          op: sum
          field: amount

Config reference

Field	Type	Required	Description
by	array	yes	Fields to group on
aggregations	map	yes	Output column name to aggregation definition

Each aggregation definition has:

Field	Type	Required	Description
op	string	yes	Aggregation function
field	string	varies	Input field to aggregate (required for all except count with *)
limit	number	no	Max items for collect
separator	string	no	Delimiter for join (default: ", ")

Aggregation functions

Function	Description	Output type
count	Count of non-null values (or all rows with *)	number
sum	Sum of numeric values	number
avg	Arithmetic mean	number
min	Minimum value	same as input
max	Maximum value	same as input
first	First value in group	same as input
last	Last value in group	same as input
collect	Collect values into a list	list
count_unique	Count of distinct values	number
join	Concatenate string values with separator	string

Examples

Revenue by category

nodes:
  revenue-summary:
    type: data.group
    config:
      by: [category]
      aggregations:
        total_revenue:
          op: sum
          field: amount
        avg_order:
          op: avg
          field: amount
        order_count:
          op: count
          field: id

Count by status

nodes:
  status-breakdown:
    type: data.group
    config:
      by: [status]
      aggregations:
        count:
          op: count
          field: "*"

Multi-field grouping

nodes:
  region-product:
    type: data.group
    config:
      by: [region, product_type]
      aggregations:
        units_sold:
          op: sum
          field: quantity
        unique_customers:
          op: count_unique
          field: customer_id

Collect and join

nodes:
  tags-per-author:
    type: data.group
    config:
      by: [author]
      aggregations:
        all_tags:
          op: collect
          field: tag
          limit: 10
        tag_list:
          op: join
          field: tag
          separator: " | "

collect gathers values into a JSON array. limit caps the array size. join concatenates values into a single string with the specified separator.

Pipeline example

nodes:
  load-sales:
    type: file.csv
    config:
      path: sales.csv

  monthly-summary:
    type: data.group
    config:
      by: [month, region]
      aggregations:
        revenue:
          op: sum
          field: amount
        deals:
          op: count
          field: id
        top_deal:
          op: max
          field: amount

  ranked:
    type: data.sort
    config:
      by:
        - field: revenue
          direction: desc

edges:
  - load-sales.output -> monthly-summary.input
  - monthly-summary.output -> ranked.input