Educational Alternatives, Volume 13, 2015

Abstract: Data mining serves an important function in providing meaningful, understandable and guiding information to various units or institutions, using data sets with a large number of instances, which are formed by numerous features that may be meaningless when dealt with separately. In this study, a questionnaire and a test were applied to 4th year primary school students on a voluntary basis and the results of the questionnaire and test were analyzed using the algorithms of Weka software, which is a data mining application tool. The study aimed to determine the factors that have an effect on student performance through the detection of the best classification algorithm for the data sets of different sizes. Following the application, the algorithms were compared according to the rate of correct estimates, Kappa statistics, F-measure and Root-Mean-Square Error. As a result, Random Forest algorithm was found out to be the most accurate classification algorithm. Moreover, the education level of the participants’ fathers and level of income were discovered to be the two factors that have the strongest effect on student performance.