ALIASED AS Minimal generalization does not guarantee minimal suppression.
Data used :
2 q ids- workclass, dob
Combinations generated : (0, 0) to (5, 5)
Concept
- Generate all the tuple combinations (0, 0) to (5, 5)
- Perform a global recoding of the entire dataset
- Check how many satisfy the constraint : here k anonymous means, need to have at least k-1 records in the same equivalence class, (no multiples of k restriction)
- Mainly the number of records that needed to be suppressed were recorded-

- As the generalisation level increases for both the qids, the number of records to be dropped reduces
- However, there is always a tradeoff between the data quality and privacy
- From a temporal attack standpoint, not very good as it can be easy to identify the records