ON PRIVACY PRESERVING DATA MINING AND FEATURE SELECTION STABILITY MEASURES: A COMPARATIVE ANALYSIS

Data mining is the mining of formerly not known and valid information from the archived data of organizations. Most of the published microdata are high dimensional due to the development in throughput technologies. It has been established that the relevant subset of features works better than full set of features. Feature selection is the dimensionality reduction technique in data mining. Selection stability is the sturdiness of the feature selection algorithms for petite perturbation of the dataset i.e., to select the same or similar subset of features in each consequent iterations. Privacy preserving in data mining refers to the area of data miming that seeks to defend privacy-sensitive information from unwanted or unsanctioned revelation and hence protecting individual data records and their privacy. Privacy preserving data mining techniques adapt the dataset for preserving the privacy of the individuals and this perturbation will influence the selection stability as it is chiefly depending on the characteristics of the dataset. There will be connection between the perturbations of the dataset for privacy preservation, feature selection stability and accuracy of the data mining results i.e., data utility. There will be diverse selection stability metrics to measure the selection stability. This paper gives different privacy preserving data mining techniques and then evaluates some of the privacy preserving data mining techniques for these different feature selection stability measures for privacy preservation, selection stability and data utility.