Group By

The Group By node rolls rows up into summaries: it collapses rows that share the same values in your grouping fields into a single output row, and computes aggregate values across each group. It is the tool for reports, totals, and analytics from detailed data.

How it works

1. Rows with matching values in the grouping fields are collected into a group.
2. Each aggregate is computed across the rows in that group.
3. The node outputs one row per group.

Input:  10 orders across 3 customers
Output: 3 rows (one per customer) with order totals

Group By reduces rows — the detail is discarded. If you need to keep the detail rows while adding a summary alongside, use the Map node's summary feature instead.

Adding a Group By

1. Drag Group By from the Functions section of the Element Panel.
2. Connect it to a data source or upstream node.
3. Open the panel to choose grouping fields and aggregates.

Grouping fields

Add one or more fields that define the groups:

• One field — Region → one row per region.
• Multiple fields — Region, Year → one row per region-and-year combination.

Rows where a grouping field is NULL are grouped together into a single "no value" group. To label them, add a Formula upstream: COALESCE(Region, "Unknown").

Aggregates

Each aggregate produces one output column. Pick a column and an aggregate function from the dropdown:

Aggregate	What it returns
Count	The number of rows in the group
Sum	The total of a numeric column
Average	The mean of a numeric column
Min	The smallest value (earliest date / lowest number)
Max	The largest value (latest date / highest number)
First	The first value seen in the group
Standard Deviation	The population standard deviation of a numeric column

The aggregates offered depend on the column's type — Average, Sum, and Standard Deviation apply to numbers only, while Count, Min, Max, and First work for any type.

How nulls and types are handled

These follow the shared strict-SQL semantics:

• Aggregates ignore NULLs — Average and Sum skip empty values rather than treating them as zero.
• A group in which every value is NULL aggregates to NULL (an empty result), not 0.
• Numeric strings are coerced where possible, so a "42" stored as text still sums correctly.

Example

Grouping: Region

Aggregates:

Aggregate	Column	Output column
Count	—	OrderCount
Sum	Revenue	TotalRevenue
Average	Revenue	AvgOrderValue
Max	OrderDate	LastOrder

Output:

Region	OrderCount	TotalRevenue	AvgOrderValue	LastOrder
East	150	45000	300	2025-03-29
West	120	52000	433	2025-03-31

Common patterns

Filter the groups (like SQL HAVING)

Aggregates can only be filtered after they are computed. Place a Filter after the Group By:

[Orders] → [Group By: CustomerID] → [Filter: TotalAmount > 10000]

Derive ratios from aggregates

Add a Formula after the Group By to compute values from the aggregate columns:

Name:       AvgOrderValue
Expression: TotalRevenue / OrderCount

Group by a date part

The grouping is by exact value, so to group by month or day, compute the key first with a Formula upstream:

Name:       YearMonth
Expression: FORMAT_DATE(OrderDate, "YYYY-MM")

Then group by YearMonth.

Tips

• Filter before grouping to reduce the rows the aggregation has to scan.
• Standardize grouping values (trim whitespace, normalize case with a Formula) so near-duplicates like "East " and "East" don't split a group.
• Preview to confirm the group count matches what you expect.

Next steps

• Expression Language — the aggregate null/type semantics
• Formula — derive metrics from aggregate columns
• Filter — keep only the groups you want
• Map — summarize a group while keeping its detail rows