Microaggregation

An anonymization technique that groups similar records into clusters of at least k members and replaces quasi-identifier values with aggregated values (median, mode). Preserves data types and statistical properties better than suppression or generalization.