A Hybrid Privacy Preserving Approach in Data Mining

Data mining algorithms extracts the unknown interesting patterns from large collection of data set. Some clandestine or secret information may be exposed as part of the data mining process. In this paper we put forward a hybrid approach for achieving privacy during the mining procedure. The first step is to sanitize the original data using a geometrical data transformation. In the second stage this sanitized data is normalized using a min-max normalization approach before publishing. For verifying the experimental results we have used kmeans clustering algorithm and it is evident from our results that this hybrid approach preserves privacy and also ensures accuracy.