Aggregations

Aggregations collapse or reshape a dataset: totals per group, a crosstab, a running total, a row count. Everything here changes the shape of the table, not just its contents.

Action	What it does	Lite
Group by	One row per group, with totals, averages or counts across the rest	●
Pivot data	Turn the values of one column into columns of their own	●
Unpivot data	Turn a set of columns back into rows	●
Count records	Add the total row count as a column	●
Window functions	Rolling, cumulative, rank, tile and partition-aggregate calculations, without collapsing rows

In Flowfile Lite

Group by, Pivot data, Unpivot data and Count records run in the browser-only Flowfile Lite build. Window functions needs the full desktop or server build.

Group by

Produces one row per combination of the columns you group by, with an aggregation applied to every other column you list.

Settings

Setting	Description
Group By Columns	Columns that define the groups.
Aggregations	One row per output column: the source column, the function, and an optional new name.
Output Column Name	Custom name for the aggregated result. Optional — defaults to the source column's name.

Aggregation functions, exactly as the drawer lists them: groupby, sum, max, mean, median, min, count, n_unique, first, last, concat. The average is mean; there is no avg.

Pivot data

Converts long data to wide: each distinct value in the pivot column becomes a column of its own, filled from the value column.

Settings

Setting	Description
Index Columns	Columns that define the rows of the final table.
Pivot Column	Its unique values become the new column names.
Value Column	The column supplying the values that fill those new columns.
Aggregations	Applied when more than one value lands in the same cell.

Pivot reads the data to discover which columns to create, so unlike most actions it runs eagerly rather than waiting for the rest of the flow.

Unpivot data

The reverse of Pivot: a set of columns collapses into two, one holding the old column name and one holding its value. This is what turns a spreadsheet laid out with a column per month into something you can group and chart.

Settings

Setting	Description
Index Columns	Columns that stay as they are, repeated once per unpivoted row.
Value Columns	The columns that collapse into name/value pairs.
Data Type Selector	Pick columns by data type instead of by name (for example, every string column).
Selection Mode	column to list columns explicitly, data_type to use the selector.

Count records

Counts the rows and returns that single number in a column named number_of_records. It has no settings.

Window functions

Adds rolling, cumulative, rank, tile or partition-aggregate columns calculated over ordered — and optionally partitioned — rows. Each function you configure produces one new column and every input row survives, which is what separates this from Group by.

Each row under Window functions is one output column: pick the function, the source column, the name to write, and the function's own parameters.

Settings

Setting	Description
Partition by	Optional. Columns that restart each calculation per group. Leave empty to calculate over the whole table.
Order by	Column(s) plus direction that define row order within each partition. Required for rolling and tile functions; not used by partition aggregates.
Window functions	One or more operations. Each takes a function, a source column, an output column name, and any function-specific parameters.

Available functions

Function	Group	Parameters	Output
Rolling sum / mean / min / max / std	Rolling	Window size in rows, and how to handle incomplete windows	Aggregate over a sliding window
Cumulative sum / count / min / max	Cumulative	—	Running total, count, min or max up to each row
Rank	Ranking	Tie-breaking method: ordinal, dense, min, max or average	Rank of each row
Tile	Ranking	Number of groups	Splits the ordered rows into N equal-sized buckets
Mean / sum / min / max / count / std / median	Partition aggregate	—	One value per partition, written to every row of that partition (SQL AVG(x) OVER (PARTITION BY g))

For rolling functions, the first rows — where the window is not yet full — can be left null (the default), computed from the partial window, or filled with 0.

Each function needs a unique output column name, and existing columns are always preserved.

Partition aggregates. With Partition by set to region, source column amount, function Mean and output name region_avg, every row gets its region's average alongside its own value:

region	amount	region_avg
north	10	20.0
north	30	20.0
south	5	5.0

This is the one-node form of a Group by followed by a Join back onto the original rows. From the Python API, df.with_columns(ff.col("amount").mean().over("region").alias("region_avg")) builds this same node.

---

← Combine Operations | Next: Output Operations →