--resample / --rolling

Two time-series reshapes, printed as a table like every other read action. The input file is never modified.

octa --resample COL --value-cols COLS FILE [--interval UNIT] [--agg FN] [--group-by COLS]
octa --rolling  COL --order-by COL --window N FILE [--agg FN] [--partition-by-cols COLS]

Both build DuckDB SQL with the same builders as the GUI Time series dialog and the resample_timeseries / rolling_window MCP tools, so all three surfaces produce identical results.

--resample: time buckets

Group rows into one bucket per interval of a timestamp column and aggregate.

Flag	Required?	Default	Description
--resample COL	yes		The timestamp column to bucket
--value-cols COLS	yes		Comma-separated columns to aggregate
--interval UNIT	no	day	minute/hour/day/week/month/quarter/year
--agg FN	no	sum	sum/mean/min/max/count/first/last
--group-by COLS	no		Comma-separated. One series per combination

The bucket lands in a column named bucket (or bucket_2 if the source already has one).

octa --resample day --interval month --value-cols amount sales.csv
octa --resample ts --interval week --agg mean --value-cols amount,qty \
     --group-by region sales.parquet

The timestamp column is cast with TRY_CAST, so one malformed value puts that row in a null bucket rather than failing the whole query.

--rolling: rolling window

Add a column holding the aggregate of the current row and the N-1 before it.

Flag	Required?	Default	Description
--rolling COL	yes		The column aggregated
--order-by COL	yes		The column that orders the frame
--window N	yes		Rows in the frame, including the current one
--agg FN	no	mean	sum/mean/min/max/count/first/last
--partition-by-cols COLS	no		Comma-separated. Restarts the frame

Output is every source column plus <COL>_rolling_<N>.

octa --rolling amount --order-by day --window 7 --agg mean sales.csv
octa --rolling amount --order-by day --window 7 \
     --partition-by-cols region sales.csv

--order-by is required and cannot be omitted: a rolling aggregate over unordered rows is meaningless.

Note the flag is --partition-by-cols, not --partition-by -- the latter is the split-into-files action.

Errors

Missing companion flags fail with exit 1 and name the flag, rather than falling back to a silent default:

$ octa --rolling amount data.csv
error: --rolling requires --order-by COL

An unknown --interval or --agg lists the accepted words.

See also

• Time Series: the GUI dialog, with a live preview.
• --sql: the same thing written by hand, when the flags do not fit.