Data reduction using metaheuristic algorithms

Abstract

Over the recent decades, the amount of data generated has been growing exponentially, the existing data mining algorithms are not feasible for processing of such huge amount of data. To solve such kind of issues, the two commonly adopted schemes are used, one is scaling up the data mining algorithms and other one is data reduction. Scaling up the data mining algorithms is not a best way, but data reduction is fairly possible. In this research work, feature selection, feature extraction and instance selection are carried out for data reduction and without downing the performance of the classification algorithms and classification accuracies. This research works conducts wide literature survey of existing instance selection, feature selection and feature extraction methods. This survey provides in depth view of existing techniques and gives a direction to rest of the study. Many approaches have been proposed for feature selection and instance selection, but most of the approaches still suffer from the problem of stagnation in local optima. Therefore, an efficient searching technique is required to address feature selection and instance selection tasks. In this research work, Cuttlefish Optimization Algorithm, Principal Component Analysis and Cuttlefish Optimization Algorithm through Tabu search are used for feature selection, feature extraction and instance selection. In the first work, instance selection is carried out by using cuttlefish optimization algorithm and the feature extraction is carried out by using principal component analysis. The obtained reduced dataset provides more or less similar detection rate, false positive rate and accuracy rate from what have obtained from using the entire dataset and it take small amount of computational time for training the classifiers newline