Time Portrait of the Student’s Behavior and Possibilities of its Use

The object of research presented in this article is the process of classifying students' behavior based on a formalized description of task performance during a certain learning cycle. The purpose of this study is to create a formalized description of the behavior of students regarding the performance of tasks during a certain learning cycle to improve the reliability of the characteristics of each student during automated data analysis in information systems. To achieve this goal, a temporal portrait of the student's behavior is proposed. This is a stylized presentation of a time series in the form of a line, the shape of which represents the delay or advance of the deadlines for completing tasks when studying a certain discipline. The main types of student behavior are highlighted. Each type of behavior corresponds to the shape of the line on the time portrait. To provide information systems with the capabilities of behavior analysis, the issue of portrait classification by neural networks has been investigated. It is proposed to perform classification using a multilayer neural network. To speed up learning and facilitate further classification, it is proposed to divide the network into several subnets, each of which can be trained independently. The issues of appropriate training of neural networks based on datasets of real training classes with groups of students are analyzed. The ability of the neural network to classify portraits of students' behavior has been proven. The results of the research can be used for data analysis in computerized learning support systems