Selectivity
The fraction of table rows matched by a predicate, a key input to cardinality estimates and the optimizer's choice between indexes, scans, and join strategies.
A predicate that matches only a few rows out of millions is highly selective; one that returns a large fraction of a table is not. Optimizers use this distribution when estimating row counts and comparing index access, full scans, and join methods.
Selectivity is not simply the number of distinct values. Skew, NULL frequency, cross-column correlation, and the actual literal or bind value can all change the result. Oracle Database and PL/SQL: architecture, SQL, and performance and the engineering meaning of indexes place it in the wider access-cost model.