A New Data Clustering Algorithm

Induction is a logical method to understand things, however, induction often can´t sufficient reflect the necessity and the regularity of things, so it needs to be complemented by deduction. The traditional clustering algorithms add the categories based on the data itself, so these approaches can be considered as the induction methods. And in order to avoid the uncertainty coming from the induction, we propose a data clustering algorithm combining inductive and deductive methods. The theory proof and experiment results show the accuracy and the ability to identify outliers, are better than some clustering algorithms.