Improving Classification Accuracy of Decision Trees for Different Abstraction Levels of Data

Abstract

Classification is an important problem in data mining. Given a database of records, each tagged with a class label, a classifier generates a concise and meaningful description for each class that can be used to classify subsequent records. Since the data is collected from disparate sources in many actual data mining environments, it is common to have data values in different abstraction levels. This article introduces the multiple abstraction level problem in decision tree classification, and proposes a method to deal with it. The proposed method adopts the notion of fuzzy relation for solving the multiple abstraction level problem. The experimental results show that the proposed method reduces classification error rates significantly when multiple abstraction levels of data are involved.