An Approach to Identifying Suspicious Student Activities During Online Programming Training Based on One-Class Classifiers

This article is dedicated to the problem of identifying suspicious student activities during online programming training in the discipline "Programming in Python" at RTU MIREA based on data on the results of students uploading solutions to exercises for mandatory tasks into the Digital Teaching Assistant (DTA) system. To solve this problem, it is proposed to apply an approach based on the development of one-class SVM classifiers trained on datasets, the values of features in which are determined taking into account the duration of solving exercises for mandatory tasks. The results of experiments on the development of one-class SVM classifiers using different kernel functions confirm the prospects of using and further developing the proposed approach to identifying suspicious student activities during online programming training.