Esra POLAT (espolat@hacettepe.edu.tr)
Department of Statistics, Faculty of Science, Hacettepe University, 06800, Ankara, Turkey
ABSTRACT
The goal of the study is finding the best data mining classification method, in determining four levels of Human Development Index (HDI) for 100 countries for the data set of 2018. Hence, Naïve Bayes, IBK, KStar, J48, RandomForest, RandomTree, REPTree, SMO, Simple Logistic, Logistic, Multilayer Perceptron methods are implemented on to the data set by WEKA data mining software for classifying 100 countries in terms of HDI levels (very high, high, medium and low) using the explanatory variables as GDP per capita, poverty rate, Net Income Gini index and Wealth Gini index. The results show that best classification method for this data set is Multilayer Perceptron with highest accuracy rate of 88%. Moreover, GDP per capita US$ is found as the most effective variable on determining the HDI levels of countries.
Keywords: Data Mining, Economic Inequality, Human Development Index, WEKA
JEL Classification: C38, I31, I32
