The authors of the article present a comprehensive analysis of the accounting of students’ academic performance in the management of the educational process of the university. The information about students that affects their academic performance and satisfaction with the educational organization is analyzed and classified. The focus of the study is on the application of predictive models in the management of the educational process in order to adapt the content of disciplines to the current contingent of students. The study used data only on first-year students (2023/24 academic year) of bachelor’s and specialist’s degree levels (n=1549). The information is depersonalized and contains the following data: demographic (age, gender, citizenship), social (socio-cultural environment, place of residence, place of residence during study), academic (previous education, results of entrance tests, current academic performance, faculty, qualification level), economic (scholarship, type of competition – budget/contract). Methods of mathematical statistics were used to analyze the data: determining the type of data distribution using the Shapiro-Wilk test, establishing the presence of multicollinearity in the construction of multiple regression by the Pearson criterion, establishing correlation dependencies by Spearman’s rank correlation method. Machine learning methods are implemented in the Python programming language (v. 3.8) using the freely distributed Keras library. The main results. The classification of factors affecting the academic performance and satisfaction of students is presented. Using the methods of mathematical statistics, the importance of each factor for predicting academic performance has been established. An educational process management model based on Agile Learning Design has been developed and presented, which allows adapting a specific discipline to the current contingent of students.
Read full abstract