Explains data profiling methods
Column profiling (null counts, distinct values, frequency and top values, minimum and maximum, patterns), cross-column and cross-table analysis, over the full population or a sample stratified so that every population is in its frame; a value with a share far above its neighbours is often a placeholder for missing (DMBOK2 ch. 13, 1.3.9 and 2.4).
Can AI do this?
People still lead Few or none of the tested AI models showed they master this. 0 of 4 models in the current benchmark master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 98% | Not shown |
| gemini-3.5-flash-lite | Low cost | 89% | Not shown |
| gpt-6-luna | Low cost | 56% | Not shown |
| claude-haiku-5-5 | Low cost | 99% | Not shown |
Mastered means a probability of at least 90% that the model has this knobit, estimated from its graded answers. “Not shown” means the answers do not settle it either way. How the models are tested.
What it looks like in practice
Expected result: Column profiling (null counts, distinct values, frequency and top values, minimum and maximum, patterns), cross-column and cross-table analysis, over the full population or a sample stratified so that every population is in its frame; a value with a share far above its neighbours is often a placeholder for missing (DMBOK2 ch. 13, 1.3.9 and 2.4).
How it is assessed: Answers a written question on the data profiling correctly
Related knobits
Needed for
- Q-402 Recognises a profiling sample excluding the problem population
- Q-404 Recognises a placeholder value counted as present
All 10 knobits of Profile and assess data quality.
Source: Practice the stage 4 reference and review checklist rest on