Classification of PISA 2012 mathematical literacy scores using Decision-Tree Method: Turkey sampling

Abstract

The purpose of this study was to classify successful and unsuccessful students in terms of mathematical literacy according to interest towards the course, attitude, motivation, perception, self-efficacy, anxiety and studying discipline variables and to determine the effect of these variables on classification. The sampling of the study consisted of the students who participated in the Program for International Student Assessment (PISA) in Turkey. Data was collected from a total of 1391 15-year-old students. CHAID analysis, which is a decision-tree technique, and data mining were used for data analysis. SPSS and WEKA software were used to analyze data. Self-efficacy perception, attitude towards the course and studying discipline were found to be the most important affective characteristics in classification of successful and unsuccessful students. It was found that accurate classification percentage obtained by J.48 decision tree, which is a data mining method, was very close to the value obtained by CHAID analysis method. These results suggest that CHAID analysis can be considered as an alternative method to decision tree methods used in data mining. According to the findings obtained from the study, firstly, self-efficacy, attitude towards the course, anxiety and studying discipline should be concentrated on in mathematical literacy for the Turkey sampling. It is believed that success status of students can be changed and Turkey can rank higher in PISA exams through the arrangements to be made in these domains.