Modeling students' academic procrastination behaviors in the context of learning analytics

In this study, it is aimed to model academic procrastination behaviors by using the traces in the online learning environment. In this context, by examining the students' homework submission behaviors; Students were grouped according to whether they made academic procrastination or not, and it was aimed to predict these groups through Moodle interaction data. 55 students enrolled in the 2020-2021 Spring Semester Database Management Systems course participated in the research. After 11 weeks of data collection, 16 features were determined. Within the scope of the first research question, the success of different classification models in predicting the student's academic procrastination behavior was investigated by using the interaction data of the students. Logistic Regression showed 90% classification accuracy with categorical variables. Within the scope of the second research question, the most important 5 attributes were determined among 16 attributes and the number of activities in the SCORM packages, the number of unique days in the system, the number of forum, quiz and homework component activities were found to be important variables in determining their academic procrastination behaviors. Finally, it was examined whether there was a difference between the academic achievements of the groups and it was seen that the students who did not procrastinate had higher end-of-term scores than the students who did. As a result, an alternative method was developed to determine students' academic procrastination behaviors in online environments, and with this method, students' procrastination behaviors were measured without intervention. This model will be important in the development of intervention methods for procrastination in future studies.