Clustering datasets consisting of numeric and nominal features is a challenging task as there are different similarity measures for numeric and nominal features. In the present paper, we propose a method to transform a mixed dataset to a numeric dataset. This method uses a similarity measure for mixed datasets and a randomly selected set of the data objects form the given mixed dataset and generate numeric similarity features. A clustering algorithm for pure numeric datasets is then applied on the newly generated numeric dataset to produce clusters. A comparative study with the other clustering algorithms demonstrated the superior performance of the proposed clustering approach. © Springer Nature Switzerland AG 2020.