Microaggregation is an important technique to the k-anonymized datasets. However, most existing microaggregation algorithms to achieving k-anonymity have some defects on distance measurement for categorical and mixed data. In this paper, we introduce a categorical data semantic hierarchy to their distance measurement to improve clustering quality. The paper also investigates mixed distance for mixed data and designs an efficient microaggregation algorithm for them. Experiments show that the distance measurement for categorical data cause less distortion, and the improved microaggregation algorithm based on the mixed distance enjoys better clustering quality than the traditional MDAV algorithm.