A data mining analysis of technology use in schools across OECD countries: The case of Pisa 2022

This study was conducted to estimate students' digital self-efficacy levels using artificial intelligence models. Based on student survey data from the OECD's PISA 2022 study, the multidimensional factors affecting self-efficacy levels in digital tasks were analyzed. The dependent variable in the study (n=1) was the ICTEFFIC index, while the independent variables (n=50) were grouped into seven sub-dimensions encompassing students' basic demographic characteristics, familiarity with information and communication technologies, creative thinking levels, social-emotional skills, school culture and climate, and well-being. Deep Learning, Decision Tree, Random Forest and Support Vector Machine models were used in the prediction process with artificial intelligence. The performances of the models were evaluated comparatively based on accuracy, precision, sensitivity and F1 score. According to the results obtained, the Random Forest model showed the highest success. The results of the analysis revealed that the most determinant variables in students' digital self-efficacy levels were mother's education level (MISCED=0.276), frequency of ICT use in lessons (ICTSUBJ=0.113), participation in creative activities (CREATAS=0.173), collaboration skills (COOPAGR=0.177), body image perception (BODYIMA=0.240) and psychosomatic symptoms (PSYCHSYM=0.177). These findings suggest that not only technological knowledge but also affective and social factors play a decisive role in digital skills. The results of the research provide guiding data for decision-makers, teachers, and educational designers regarding the development of digital competencies.