The article discusses approaches to the problem of increasing academic performance of the university students. Academic performance indicates the percentage of positive ratings received by students as a result of the exam sessions, excluding retakes. Analyzing factors affecting the academic performance is seen on basis of information from the assessments database of UMU portal of Don State Technical University. Highlighted basic factors (objects) that affect academic performance: study groups, teachers, subjects, exams timetable. In this paper, application of clustering methods allowed to use variance analysis of averages. Through two-way analysis of variance (ANOVA) established significant impact of object classes "Teacher" and "Subject" on the percentage of positive evaluations received by students as a result of the session. Given an example of group, teachers and subjects classes, on which the examinations timetable has a significant impact on exam results

Keywords: higher education, academic performance, ANOVA, factors, clustering