Predicting Mathematics Performance by ICT Variables in PISA 2018 through Decision Tree Algorithm

Abstract

Considering the large volume of PISA data, it is expected that data mining will often be assisted in making PISA data more meaningful. Studies show that different dimensions of ICT may reveal different relationships for mathematics achievement. The purpose of this article is to evaluate the success of the decision tree classification algorithms in predicting the effect of ICT on students' mathematics performance. The population of the research consists of 15-year-old students studying in Turkey. The sample of the study consists of 6570 students who participated from Turkey and gave adequate answers to the ICT Familiarity Questionnaire in PISA and whose mathematics score was calculated. The J48 algorithm is more successful in classifying students with low mathematics achievement than classifying students with high mathematics achievement. The rate of correctly predicting mathematics achievement with weighted average values and variables related to ICT is 66.1%. ENTUSE [ICT use outside of school (leisure)], ICTCLASS [Subject-related ICT use during lessons] and USESCH [Use of ICT at school in general] variables are the most effective variables. It is thought that the reason for the difference in the effect of the use of information and communication technologies for entertainment purposes on mathematics achievement is since the level of recreational use can have a positive effect up to a certain level, while excessive use can be harmful.