A Comparative Analysis on Evaluation of Classification Algorithms Based on Ionospheric Data

Abstract

Data mining technique is an application of the regular process for analyzing the huge size of existing data, excavating valuable information to support the decision-making process. The Earth’s upper atmosphere consists of an ionized part referred to as the ionosphere. It lies between eighty kilometre to one thousand kilometer height above the sea level, an area which comprises the parts of the thermosphere, mesosphere as well as the exosphere. The ionosphere is a shell of electrons and electrically stimulated atoms that ambiances the Earth. The target for Weka tool classification are these free electrons in the ionosphere. The performance analysis and experimental results carried out for five classifiers such as Naive Bayes, SVM, ANN, K-NN, and J48 are compared and evaluated in this study. The overall performance of these algorithms is analyzed based on the classification accuracy in which decision tree algorithm has achieved best performance compared to other algorithms. The above accuracy in ionospheric data classification is the focal idea of assessing the performance in data mining algorithms.