Privacy Preserving Fuzzy Based Decision Tree Classifier

Abstract

Huge volumes of personal data are collected and sharing of these data is beneficial for machine learning and data mining applications. These data are very much important for decision making in the competitive business world. Therefore, privacy preserving has been developed to hide sensitive data. We present a privacy preserving fuzzy based decision tree classifier, which allows extracting rules from the decision tree and the sensitive data are fuzzified in order to preserve the privacy. The objective of privacy preserving data classification is to build accurate classifier without revealing private information in the data being mined. In this paper, the advantages of fuzzy logic and decision tree are used to build the privacy preserving classifier. Fuzzy logic is used to preserve the privacy in the system and decision tree is used for classification. The effectiveness of the classifier is verified by experiments on Hepatitis dataset provided by the University of California, Irvine (UCI) machine learning repository. The classification accuracy obtained using this approach is 83.63%.