In recent years, the internet has changed modern life in many ways. Internet technologies have transformed the whole field of education, health, defense industry, and health into a new format. However, the internet has changed social life and human-human, human-machine interaction, and these interactions have created addictions at various levels. Internet addiction is accepted as a current and serious disease that can affect mental health. The aim of this study is to model the factors affecting the addiction levels of students from two countries (Türkiye, Iraq) using the Multivariate Adaptive Regression Extensions method. The primary purpose of this study is to monitor the performance of MARS, which is one of the multivariate statistical methods, on a data set and to test its predictive ability on the data types used in the research. On other hands, the main purpose of the study is to test the applicability of MARS on the variable types and data characteristics used in the research and to reveal the effectiveness of this method to the researchers who will work with this type of data. The secondary aim of the study is to model the factors that trigger internet addiction, which is seen as an important disease and danger of this age, through some measurement tools. Here, it was also desired to examine whether the factors affecting internet addiction differ between cultures. Therefore, the sample of the study consisted of university students from Türkiye and Iraq.

The MARS method used in the thesis is a non-parametric data mining technique. MARS puts a knot at each point where linearity will end in the relationship between the variables and creates appropriate functions for each linear extension it obtains. In this respect, it can model the cause-effect relationship in a more rational space.

The data set for the thesis study consisted of the internet addiction Scale and the data obtained from the questionnaires containing demographic/personal information. The sample of the study consists of 2235 students, 1220 (427 boys and 793 girls) from Türkiye and 1015 (465 boys and 550 girls) from Iraq, using the random sampling method. The data within the scope of the study were benefited from the SPM 8.2 program. As a result of MARS analysis; while 9 (nine) basic functions were obtained for students in Türkiye, the number of basic functions was observed as 20 (twenty) for the data set consisting of Iraq. In the model created by MARS; It has been determined that the most important factor affecting the internet addiction level of both Turkish and Iraqi students is “daily internet usage time”. The details of the study are in the Findings and Conclusion sections of the thesis. We hope that the findings on the factors affecting the level of internet addiction with the MARS method will contribute to the literature, and it is recommended that such data mining methods be applied to life-oriented data together with the theory.