Group a typed DataFrame by one or more key columns.

grouped = groupBy @'["department"] employees

Build a named aggregation entry. The result column name is supplied via TypeApplications; the underlying expression is validated against the source schema at compile time.

as produces a transformer on the aggregation chain — entries compose with plain (.) from Prelude (or via (|>) for SQL-like postfix reading). aggregate applies the composed transformer to the empty chain internally, so no terminator is needed.

Prefix form

result = grouped |> aggregate
    ( as @"total"  (sum   (col @"amount"))
    . as @"orders" (count (col @"order_id"))
    . as @"avg"    (mean  (col @"amount"))
    )

Postfix form (SQL-like)

result = grouped |> aggregate
    ( (sum   (col @"amount")   |> as @"total")
    . (count (col @"order_id") |> as @"orders")
    . (mean  (col @"amount")   |> as @"avg")
    )

Run a typed aggregation against a grouped DataFrame.

The first argument is a chain of as entries composed with (.). The empty composition (id) yields just the group keys. The result schema is the group-key columns followed by the aggregation columns in declaration order.

result = grouped |> aggregate
    ( as @"total"  (sum (col @"amount"))
    . as @"orders" (count (col @"order_id"))
    )
-- result :: TypedDataFrame
--     '[ '("region", Text)
--      , '("total", Double)
--      , '("orders", Int)
--      ]

Escape hatch: run an untyped aggregation and return a raw DataFrame.