Discovery of characteristic patterns from tabular structured data including missing values

Abstract

This paper proposes a method dealing with missing values in the discovery of frequent patterns. Also, it proposes a method that effectively discovers the patterns from examples composed of attributes and their values. The method generates candidate patterns based on the combination of attributes and the combination of attribute values. It evaluates the patterns based on two supports. These supports are calculated based on the number of examples that do not include missing values in attributes composing target items and in all attributes. The proposed method is verified by comparing it with the existing methods dealing with missing values.