Summarize
Group by one or more columns and compute aggregate values: sum, average, count, median and more.
What it does
Rolls rows up into summary rows. Pick the Group-by columns (one output row per distinct combination), then add aggregations: for a column, a function (Sum, Count, Count Non-Null, Count Distinct, Min, Max, Average, Median, Std Dev, Variance, First, Last, Concatenate, or Percentile) that reduces each group's values to one cell. With no group-by columns the whole table collapses to a single grand-total row. Each aggregation gets its own output column, named however you like.
Inputs and outputs
| Inputs | One. |
| Outputs | One. |
When to use it
- Totals or averages by category: sales per region, orders per customer
- Counts per group, including a clean Count vs Count Non-Null vs Count Distinct split
- Reduce a large table to a compact summary before a join or output
- A grand total (leave Group by empty): one row of totals for the whole table
Examples
Sales by region
Group by: region · Sum of amount
One row per region; column Sum_amount.
Multiple stats at once
Group by: region · Sum of amount · Average of amount · Count
Aggregate the same column many ways.
Conditional total
Sum of amount · Only when [status] = 'paid'
Counts only the rows that match, with no pre-filter node needed.
First & last
Group by: user · First of event · Last of event
Follows the input row order, deterministically.
Tips
- Name each output column, or leave it blank for a sensible default (Sum_amount, Count, …).
- Numeric-only functions (Sum, Average, Median, Std Dev, Variance, Percentile) are grayed out for text columns; Min/Max/First/Last/Count work on anything.
- First, Last and Concatenate follow the input row order; add a Sort upstream to control it.
- Concatenate joins the group's values with a separator; turn on “Only distinct values” to drop repeats.
- Percentile is interpolated: 50 is the median, 90 is P90.
- Output rows come back sorted by the group-by keys, so runs are reproducible.