Examination of reading literacy levels in PISA 2018 Turkey sample with different data mining classification methods

The purpose of research is to determine the classification accuracy of students' success status and reading skills proficiency levels according to the factors affecting the success of students' reading skills and their success scores based on the PISA 2018 Turkey sample by using Artificial Neural Networks, Decision Trees, K-Nearest Neighborhood and Naive Bayes methods and to examine the general characteristics of success groups. In the research, 6890 student questionnaires were used. Firstly, the missing data were examined and completed. Secondly, 24 index variables were determined by examining the literature, PISA 2018 Technical Report and data. Thirdly, the students were scaled in 2 categories as "Successful-Unsuccessful" according to the scores of PISA 2018 reading test and in 3 categories as "Level-1", "Level-2" and "Level-3" according to their proficiency levels. Statistical analysis was conducted with SPSS MODELER. At the end of the research, Decision Trees C5.0 had the highest classification rate with 89.6%, QUEST had the lowest classification rate with 75%, and four clusters were obtained with the Two-Step Clustering analysis method to according to the success scores. C5.0 had the highest classification rate with 88.6% and the QUEST had the lowest classification rate with 61.7%, and three clusters whose distributions are not proportionally close to each other were obtained. It can be said that the data sets are suitable for clustering and according to both their achievement scores and their levels, all data mining methods can be used to classify students because of their ability to correctly classify beyond random classification.