Classification of Categorical Uncertain Data Using Decision Tree

Abstract

Certain data is a data whose values are known precisely whereas uncertain data means whose value are not known precisely. But data is always uncertain in real life applications. In data uncertainty attribute value is represented by a set of values. There are two types of attributes in data sets namely, numerical and categorical attributes. Data uncertainty can arise in both numerical and categorical attributes. Traditional decision tree algorithms work with certain data only. The classification performance of decision tree can be improved if complete information of data is considered. Probability Density Function (PDF) is used to improve the accuracy of decision tree classifier. Existing system to handle uncertain data works on only numerical attributes means only range of values. They cannot works uncertain categorical attributes. This paper proposes a method for handling data uncertainty on categorical attributes. The decision tree algorithm is extended to handle uncertain data. The experiments show that the classification performance of this decision tree can be enhanced.
DOI: 10.17762/ijritcc2321-8169.150661