In this study, the factors affecting the transactional distance levels of university students who continue their courses with distance education in the 2020-2021 academic years due to the COVID pandemic process were examined. Factors that affect transactional distance are modeled with Artificial Neural Networks, one of the data mining methods. Research data were collected from a total of 1638 students, 546 males and 1092 females, studying at various universities in Turkey, by using the personal information form, the Transactional Distance Scale and the Social Anxiety Scale in E-Learning Environments. Students' transactional distance levels were included in the model as dependent variable and social anxiety and 17 variables, which were thought to be theoretically related to transactional distance, were included in the model as independent variables. The research data were analyzed using Multilayer Perceptron (MLP) Artificial Neural Networks and Radial Based Functions (RBF) Artificial Neural Networks methods. In addition, these methods are compared in terms of estimation performance. According to the results of the research, it has been seen that the MLP method predicts the model with lower errors than the RBF method. For this reason, the results of the MLP were taken into account in the study. As a result of the analyzes carried out with this method, quickness of the instructor to give feedback on messages is determined as the most effective variable on the transactional distance.