Methods of classifying the distance learning system users in the model of constructing their personalized learning strategies

The article examines the problem of adaptation of the distance learning system to the contingent of users by constructing personalized learning strategies using their previous tests results. The main part of the suggested model is classifying users by various academic progress criteria. Comparative analysis of results of applying different classifiers for this purpose is presented. The following types of classifiers were used: Bayes classifier, logistic regression, k-nearest neighbors algorithm, decision tree, random forest, boosting, and bootstrap aggregating classifier that uses a majority vote as the voting scheme. The article presents the results of a numerical experiment using the data on the work of MAI distance learning system CLASS.NET.