Feature Selection using Clustering Algorithms: FAST and LUFS

Feature selection is used to reduce the number of features in many applications where hundreds or thousands of features are present in data.  Many feature selection methods are proposed which mainly focus on ï¬nding relevant features. High dimensional data becomes very common with the emerging growth of   applications. Thus, there is a need of mining High dimensional data very effectively and efficiently. Clustering is widely used data mining model that partitions data into a set of groups, each of which is called a cluster. To reduce the dimensionality of the data and  to select a subset of useful features from this clusters is the main goal of feature subset selection.  In dealing with high-dimensional data for efficient data mining, feature selection has been shown very effective. Popular social media data nowadays increasingly presents new challenges to feature selection. Social media data consists of data such as posts, comments, images, tweets, and linked data which describes the relationships between users of social media and  the users who post the posts. The nature of social media increases the already challenging problem of feature selection because the social media data is massive, noisy, and incomplete. There are several algorithms applied to find the efficiency and effectiveness of the features. Here we are using the combination of FAST and Linked Unsupervised feature selection algorithm for the linked high dimensional data.