A New Privacy Preserving Data Mining Algorithm with better Feature Selection Stability and Accuracy

ABSTRACT Data mining digs up formerly not known and precious type of patterns and information acquired from hefty storage of data that is archived. In the latest couple of decades, those progressions over internet technologies realize colossal augmentation in the dimensionality of the dataset stressed for data mining. Feature selection is a vital dimensionality reduction method as it advances accuracy, efficiency and model interpretability of data mining algorithms. Feature selection stability may be viewed to be the sturdiness of the algorithm for feature selection which facilitates picking alike or the same subset of features for minute perturbations in the dataset. The indispensable inspiration driving data mining that is utilised for the privacy preservation is the adaptation of original datasets by means of a method to safeguard privacy of the persons and work out consequent data mining algorithm to acquire information from it. This has led to pull together users' individual data and fed into data mining systems which should guarantee that there is no trouncing of privacy. This perturbation of the dataset will influence the feature selection stability. There will be a relationship sandwiched between privacy preserving data mining and feature selection stability. This paper scrutinizes on this issue after giving an introduction to privacy preserving data mining methods and also it brings is a new-fangled privacy preserving algorithm which has less impact on feature selection stability in addition to accuracy. Keywords: Data Mining, Privacy Preservation, Feature Selection, Selection Stability, Kuncheva Index