concat

concat stacks peer tables row-wise and adds a label column that records where each row came from.

It is most useful when two or more branches represent comparable cohorts, batches, sites, or time slices and you want one tidy table downstream.

Stacking contract

Field	Required	Notes
inputs	yes	Two or more upstream tables.
groupColumn	yes	New column added to every output row identifying the source input.
groupLabels	no	Labels for each input; defaults to the input refs.
schemaMode	no	strict by default; intersect keeps shared columns, union fills missing cells with null.

Schema mode is the decision

• strict requires the same column set and is safest when tables should match exactly.
• intersect keeps only shared columns and can silently drop useful fields if you are not looking.
• union keeps all columns and fills missing cells with null, which is flexible but should be followed by null-profile review.

Result shape

default is the combined table with the added groupColumn. Check that the labels are readable, because those values often become filters or group names later.

Example

- id: combined
  kind: concat
  inputs: [batch_a, batch_b]     # ≥2
  groupColumn: batch              # added column tagging each row's source
  groupLabels: [a, b]             # optional; defaults to input ref strings
  schemaMode: union               # strict | intersect | union, default strict

Related

• t_test - grouped tests often start by concatenating two cohorts