Transformations

Transformations are a no-code layer that operates on your published dataset results. Instead of editing the SQL, you point and click to reshape the data, group, aggregate, filter, sort, hide columns, and limit rows. Transformations run in-browser using DuckDB, so they are instant and don’t touch your database.

What you can do

Group by

Group rows by one or more columns to create aggregated views. When you enable grouping on a column, other columns need an aggregation function.

Aggregate

Apply aggregation functions to columns. Available operations depend on the column type:

Category	Operations
All types	Count, Count values, Count empty, Count unique, Percent empty, Percent non-empty
Numeric	Sum, Average, Median, Min, Max, Variance
Boolean	Count true, Count false, Percent true, Percent false
Text	First, Last, Concat
Date/time	Earliest, Latest

Filter

Apply conditions to narrow the rows displayed. Filters support a wide range of modes:

Category	Modes
Text	Matches, Contains, Not contains, Starts with, Ends with, Empty, Not empty, In, Not in
Numeric	Equal, Not equal, Greater than, Greater or equal, Less than, Less or equal, Between
Boolean	Is true, Is false
Date	Equal, Between, Relative (previous/current/next N hours/days/weeks/months/quarters/years)

Multiple filters can be combined with AND or OR logic.

Sort

Sort by any column in ascending or descending order. You can stack multiple sort criteria, they apply in priority order.

Hide columns

Toggle column visibility to hide columns you don’t need in the current visualization. Hidden columns are still available if you re-enable them later.

Row limit

Cap the number of rows returned. Useful for top-N views or limiting large datasets.

Where transformations are used

Dataset results panel: shape how you see your data in the table while exploring a dataset.
Charts on dashboards: each chart has its own transformation configuration, so the same dataset can power multiple charts with different groupings and filters.
Chart-specific transformations: graphs can have an additional transformation layer for shaping data specifically for the visualization. See Chart transformations.