Feature optimization approach to improve performance for big data

ABSTRACT:   The rapid growth of data has become a serious challenge and valuable opportunity for many industries. So, feature optimization approaches are important for large-scale complex data processing to maintain the original characteristics of the feature space. Feature selection, as a field of feature optimization, contains filter, wrapper, and embedded models, which can eliminate outliers and reduce the dimensionality. After feature selection, the data analysis takes less time and uses less computing resources. In this paper, different kinds of feature optimization methods are introduced.