gpt-6-astra
gpt-6-astra masters 152 of the 199 benchmarked knobits of Data modelling: an expected 90% of the competence.
Data modelling, activity by activity
| Activity | Test score | Knobits mastered (expected) |
|---|---|---|
| Capture data requirements | 75% | 78% |
| Define business terms | 92% | 94% |
| Document data rules as SBVR fact types and rule statements | 89% | 94% |
| Build the conceptual model and taxonomies | 88% | 87% |
| Derive the logical data model | 89% | 91% |
| Design the physical data model | 91% | 97% |
| Derive data quality rules | 94% | 98% |
| Generate and review DDL | 74% | 81% |
Where it falls short
The knobits with the lowest probability of mastery: review its output here most carefully.
- K-133 Recognises a missing or wrong requirement trace 0%
- K-203 Recognises a negative definition 0%
- K-405 Recognises a node without warrant 0%
- K-309 Recognises a gap in a partition of cases 6%
- K-821 Recognises a restricted column without access design 19%
- K-117 Recognises a missing or wrong requirement type 20%
- K-810 Recognises a truncated identifier taken as the real name 20%
- K-828 Recognises a business date stored with a time of day 20%
- K-519 Recognises undeclared subtype constraints 35%
- K-121 Recognises duplicate data requirements 38%
- K-122 Recognises paraphrased data requirements 39%
- K-123 Recognises conflicting data requirements 39%
- K-808 Recognises a CHECK constraint parsed and ignored 40%
- K-116 Recognises a missing or meaningless requirement priority 40%
- K-817 Recognises a length taken from sample data 41%