A Data-Driven Approach to Determining Safe Classroom Capacities During the Transition to Face-to-Face Education

In this paper, different models have been developed to estimate how many students should be in the existing classrooms to be less affected and protected from the Covid19 virus during transition to face-to-face education. The factor that determines the risk of transmission of the Covid 19 virus is not only physical distance, but the duration of exposure. In this direction, model has been created by Fuzzy Logic method to evaluate the efficiency of classrooms in terms of physical sizes using the classroom and window sizes of existing primary schools. Various models have been developed by using the data obtained in line with the developed model. After the evaluation of the obtained models, it was concluded that deep neural networks model can be accepted as a more suitable approach for this estimation problem than other supervised learning methods. It is expected that the developed model will help the guidelines prepared for taking necessary precautions in educational structures and making arrangements to prevent the transmission of the virus. Developed with the data obtained by examining only the primary school classrooms, developed models can also be applied with the data to be obtained by examining the classrooms of different levels.