Sampling and analyzing statistical data to predict the performance of MOOC

MOOC platforms allow to accumulate a large amount of statistical data on various activities of course participants. This provides an opportunity for predicting the performance of an online course even at a stage when the course is not completed, and its students have time to correct their situation. To make a forecast, the first priority task is to create the correct sample for supervised learning models. The hypothesis that practical exercises that are performed in the first half of the course have a significant impact on the performance of the course, and among them, the most laborious in time and effort is the most influenced, received experimental confirmation. In the online course “Methods and Algorithms of Graph Theory”, Spearman’s correlation for Problem 6 using the Magu-Weismann algorithm is the highest (0.58 and 0.59).