Compute aggregations (mean, median, sum, min, max, count, count_distinct, std, var, percentile, mode) over a table. With no group_by columns, returns a single-row table; with grouping columns, returns one row per group. Accepts CSV or Parquet input and emits Parquet output.
JSON list of aggregation specs. Each spec is an object with keys column, function (mean, median, mode, sum, min, max, count, count_distinct, std, var, percentile), optional p (for percentile), and optional as (alias).
group_by
string
Comma-separated grouping columns. Empty (default) aggregates the whole table into a single row.