Data Transformations
Transformations let you reshape, filter, and enrich your data before it reaches a chart or table. You can chain multiple transformations together in a pipeline — each one takes the output of the previous step as its input, so the order matters.
How the pipeline runs
Transformations attached to a panel run as an ordered list. Each transformer has:
- a type (e.g.
DF_SELECT_FIELDS) and a config object, - an enabled flag — toggle a step off without deleting it,
- an order — lower numbers run first.
Dashboard-level global transforms run before a panel's own pipeline, so they're a good place for cleanup that every panel needs.
Two kinds of transform
| Stage | Works on | Types |
|---|---|---|
| Raw JSON | The API response as-is, before any table conversion. | EXTRACT, SELECT, RENAME, FIELDS, FILTER, SORT, LIMIT, REDUCE |
| DataFrame | The data after it's been turned into a table of columns. | every DF_* type |
Raw-JSON transforms are ideal for navigating nested JSON and trimming the response down. Once your data is tabular, the DF_* transforms take over for filtering, calculating, grouping, joining, and reshaping.
Tip: while building a pipeline, set the panel's chart type to JSON to watch how each step changes the data.