Performance Evaluation of Anonymized Data Stream Classifiers
Aradhana Nyati, Divya Bhatnagar · E-LIS Repository (University of Naples Federico II) · 2016
Data stream is a continuous and changing sequence of data that continuously arrive at a system to store or process. It is vital to find out useful information from large enormous amount of data streams generated from different applications viz. organization record, call center record, sensor data, network traffic, web searches etc. Privacy preserving data mining techniques allow generation of data for mining and preserve the private information of the individuals. In this paper, classification algorithms were applied on original data set as well as privacy preserved data set. Results were compared to evaluate the performance of various classification algorithms on the data streams that had been privacy preserved using anonymization techniques. The paper proposes an effective approach for classification of anonymized data streams. Intensive experiments were performed using appropriate data mining and anonymization tools. Experimental result shows that the proposed approach improves accuracy of classification and increases the utility, i.e. accuracy of classification while minimizing the mean absolute error. The proposed work presents the anonymization technique effective in terms of information loss and the classifiers efficient in terms of response time anddata usability.