Feature Selection Is Important: State-of-the-Art Methods and Application Domains of Feature Selection on High-Dimensional Data

With the advancement of technologies in the big data field, feature selection plays a vital role in most of the prediction problems and many application domains including healthcare, government sectors, network attacks prediction, microarray data analysis, etc. Nowadays, due to the existence of enormous volume of data with high-dimensional attributes and data types, it has led to a problem to find and classify informative features from noninformative ones. To solve these issues, filter, wrapper, embedded, and hybrid methods are used. In this chapter, we provide a detailed introduction about the feature selection with recent state-of-the-art techniques with respect to filter, wrapper, embedded, and hybrid models and discuss taxonomy of the dimensionality reduction techniques and fuzzy logic-based feature selection techniques. Further, we have given importance to feature selection among various application domains such as text analytics, video analytics, audio analytics, microarray analysis, intrusion detection systems, and feature selection in stream data analysis. Finally, we conclude by explaining application domains of feature selection with elaborate discussions.