We need a new Datalab issue type, identifier column , to automatically flag columns that are likely to be…
We need a new Datalab issue type, identifier column , to automatically flag columns that are likely to be…: a task in LegoFlow-SWE (Harbor dataset). A column consisting entirely of consecutive integers (possibly shuffled) is almost certainly an identifier column. Including it as a feature can…
The task
A column consisting entirely of consecutive integers (possibly shuffled) is almost certainly an identifier column. Including it as a feature can cause models to overfit or learn spurious patterns. Currently, Datalab has no built‑in check for such columns.
Part of Lego-X/LegoFlow-SWE.