The presented study makes it possible to expand the list of characteristics used in the ranking of universities. The material for the study was the results of monitoring in the field of e-learning of federal and national research universities, according to the results of diagnostics of the development of e-learning conducted by the Ministry of Education of the Russian Federation in 2013. Projects that are being implemented in Russia in the field of e-learning provide an opportunity to identify leaders in this area in various kinds of ratings. The standard procedure for calculating the arithmetic mean values of several quantitative assessment criteria in fact leads to the loss of valuable individual information on the activities of each of the rating participants. In this article, a fundamentally different approach based on the principles of cluster analysis is proposed. It is based on the fact that any set of externally homogeneous objects tends to differentiate. The formation of clusters is due to the proximity of the values of particular criteria used in the construction of the rating of each participant in the assessment. The article uses the method of average cluster analysis, while analyzing its results on the sample under study, the elements of which are the values of individual evaluation criteria for universities, with a different number of clusters. For each fixed number, the degree of differentiation and the type of broken lines for each of the clusters is determined. It has been established that when dividing the set of federal universities into two clusters, their isolation is pronounced. The formation of the three clusters led to the fallout into the third cluster of only one of the federal universities. Therefore, a further increase in the number of clusters was considered inexpedient. Thus, the ranking of federal universities has acquired a new context the division of two echelons into universities, with practically non-intersecting lines of averages. When clustering was carried out on a single set of universities, a pattern of differentiation was observed only by the levels of values of individual criteria that do not correlate with the status of the university. This circumstance indicates the success of functioning in the field of e-learning for universities of both types. Thus, the proposed method of cluster rating construction allows, in contrast to existing approaches, to most effectively assess the work of universities of various types in the field of e-learning on the basis of selected private criteria. Such a differentiation will allow to highlight and further spread the best practices in this field of education.
