Neural networks in econometric modeling of assessment of the quality of the educational process at a university

This article discusses the development and application of econometric neural network models for assessing student performance as a criterion for the quality of education received. The information base for the study was based on the data of a survey conducted among students of the economics, physics, mathematics and electrical power engineering faculties, as well as data on expelled students presented by the deans of the Faculty of Economics and the Faculty of Mathematical Sciences of the Institute of Digital Technologies of the Mari State University. As part of the study, two econometric neural network models were considered. In the first, the dependent variable is the “Number of academic debts” indicator, and in the second - the “Number of absences” indicator. For each of the models, five types of neural networks were built. The main characteristics of the networks were assessed and the best models were selected. To substantiate the practical significance, the work provides examples of the use of the constructed models.