An Efficient Microaggregation Method for Protecting Mixed Data

MDAV2k is an efficient multivariate data-oriented microaggregation technique for microdata protection. However, the method works only on numeric data thus prohibiting it from being used to protect real world data containing mixed values. In this paper we introduce an improved data-oriented microaggregation method for protecting microdata in the case that the records under consideration are expressed in terms of continuous and categorical attribute values. Experimental results shows that the MDAV2k method with improved mixed distance measurement can microaggregate mixed data to achieve better k-anonymity with less information loss and a better trade-off between information loss and data disclosure risk.