Feature Selection: A Review

In everyday life, we use a large number of web applications which result in the storage of a huge volume of data both column-wise and row-wise. Along with the explosion of web applications, the emergence of IoT resulted in the creation of a mammoth volume of dynamic data. This may result in the inevitable use of dimensionality-reduction techniques which is a vital part of the preprocessing of data for data mining applications. The mostly used dimensionality-lesseningmethods are feature selection and feature extraction, of which feature selection is superior in functionality. The heftiness of feature selection algorithms. for a slight amendment of data is called feature selection stability. Selection stability is deliberated as one of the significant criteria of feature selection algorithms along with data utility. The significance of feature selection stability is very much considered for dynamic data like incremental micro data as well as privacy-preserved micro data, as feature selection stability is mostly data reliant.