Privacy preserving sub-feature selection in distributed data mining

•It selects sub-feature data from data miner database.•It preserves the privacy of each individual party's data during collaborative computation.•The model collects the correct number of sub-feature data from different data set.•Fuzzy random variables help to select sub-feature data for generating special class.

This paper addresses the selection of sub-feature from each feature using fuzzy methodologies maintaining the privacy during collection of data from participating parties in distributed environment. Based on fuzzy random variables conditional expectation is used in which two fuzzy sets are generated using Borel set that helps to determine sub-feature within certain interval. The privacy and selection of sub-feature leading to a distinguished class is the main objective of this research work. These two problems are directly related to data mining problems of classification and characterization of feature. In many cases traditional techniques are not suitable for complex databases. However our methodology provides better way for selection of sub-features under different situations. The proposed model and techniques both presents extensive theoretical analysis and experimental results. The experiments show the effectiveness and performance based on real world data set.