Review on OFS: Online Feature Selection based on Regression analysis and Clustering method along with its Application

Abstract

In Data mining the Feature selection is one of the main techniques. In this its result shows, almost all learning of feature selection is finite to batch learning. Not similar to existing batch learning methods, online learning can be chosen by an encouraging family of well-organized and scalable machine learning algorithms for large-scale approach. The large scale quantity of online learning needs to retrieve all the features/attributes of occurrence. The difficulty in Online Feature Selection in which the online learner is allowed to maintain a classifier that involved a small and fixed or exact number of features. This article demonstrates two different tasks of online feature selection. First one is learning with full input and second is learning with partial input. The sparsity regularization and truncation techniques are used for developing the algorithms. There is a challenge of online feature selection is how to make prediction accurately for an instance using a small number of active features in high dimensionality. The proposed system presents novel method such as Multiclass classification, Regression analysis and Clustering method to clear up each of the two problems and give their performance analysis.