Clustering based Feature Selection from High Dimensional Data

Data mining techniques have been widely applied to extract knowledge from large databases. Data mining searches for relationships and global patterns that exist in large databases that are ‘hidden’ among the huge data. Feature selection involves selecting the most useful features from the given data set and reduces dimensionality. Graph clustering method is used for feature selection. Features which are most relevant to the target class and independent of other are selected from the cluster. The feature subset obtained are given to the various supervised learning algorithms to increase the learning accuracy and obtain best feature subset. The feature selection can be efficient and effective using clustering approach. Based on the criteria of efficiency in terms of time complexity and effectiveness in terms of quality of data, useful features from the big data can be selected. DOI: 10.17762/ijritcc2321-8169.15061