Pragmatic Investigation on Performance of Instance-Based Learning Algorithms

Abstract

In the recent past , there has been increasing usage of machine learning algorithms for classifying the data for various real-world applications such as weather prediction, medical diagnosis, network intrusion detection, software fault detection, etc. The instance-based learning algorithms play a major role in the classification process since they do not learn the data until the need of the developing the classification model. Therefore, these learning algorithms are called as lazy learning algorithms and implemented in various applications. However, there is a pressing need among the researchers to analyze the performance of various types of the instance-based classifier. Therefore, this chapter presents a pragmatic investigation on performance of the instance-based learning algorithms. In order to conduct this analysis, different instance-based classifier namely instance-based one (IB1), instance-based K (IBK), Kstar, lazy learning of Bayesian rules (LBR), locally weighted learning (LWL) are adopted. The performance of these algorithms is evaluated in terms of sensitivity, specificity, and accuracy on various real-world datasets.