Problem. In this article, the student behavior models are formalized, considering their motivation to improve the performance. The relevance of this topic is due to the need to introduce new information technologies into the education system of Ukraine, which are aimed at reforming the educational process in order to overcome the crisis in education and increase the level of professional education of university graduates. Goal. In the process of developing models, a large number of works were considered in which the problems of mathematical modeling and management of the learning process were discussed, with the aim of increasing student performance. The purpose of this article is to describe the behavior patterns of students for the development of information technologies aimed at improving academic performance. Methodology. To achieve this goal, the methods of cluster analysis, fuzzy sets and production modeling were used. Clustering parameters were determined as a result of processing statistical data obtained experimentally. According to the results of the experiments, the students were divided into typological groups depending on their performance, for this purpose a modified cluster analysis method was used. Considering poorly formalized factors, this task was presented as a decision-making task under conditions of uncertainty. The result of clustering was presented with the help of a fuzzy function, and production models of students' behavior during the test (exam) were built. Results. As a result, 5 models of students' behavior during the test (exam) were identified and described, and these models were formalized in the language of predicates. Matrices of fuzzy relations of students' preferences for evaluating their knowledge are built and formalized. The relations of strict preference for each model of student behavior are defined and a number of clearly non-dominant alternatives of student behavior that want to improve their academic performance are identified. Graphic representations of fuzzy possibilities of redistributing students to another typological group by academic performance, depending on their behavioral patterns, are constructed. Originality. This article proposes the production models of building up student behavior during the test (exam). Practical value. Formalization of student distribution models by typological groups depending on academic performance can be used in the development of an automated tool for creating information technology to form individual methods of teaching studentsProblem. In this article, the student behavior models are formalized, considering their motivation to improve the performance. The relevance of this topic is due to the need to introduce new information technologies into the education system of Ukraine, which are aimed at reforming the educational process in order to overcome the crisis in education and increase the level of professional education of university graduates. Goal. In the process of developing models, a large number of works were considered in which the problems of mathematical modeling and management of the learning process were discussed, with the aim of increasing student performance. The purpose of this article is to describe the behavior patterns of students for the development of information technologies aimed at improving academic performance. Methodology. To achieve this goal, the methods of cluster analysis, fuzzy sets and production modeling were used. Clustering parameters were determined as a result of processing statistical data obtained experimentally. According to the results of the experiments, the students were divided into typological groups depending on their performance, for this purpose a modified cluster analysis method was used. Considering poorly formalized factors, this task was presented as a decision-making task under conditions of uncertainty. The result of clustering was presented with the help of a fuzzy function, and production models of students' behavior during the test (exam) were built. Results. As a result, 5 models of students' behavior during the test (exam) were identified and described, and these models were formalized in the language of predicates. Matrices of fuzzy relations of students' preferences for evaluating their knowledge are built and formalized. The relations of strict preference for each model of student behavior are defined and a number of clearly non-dominant alternatives of student behavior that want to improve their academic performance are identified. Graphic representations of fuzzy possibilities of redistributing students to another typological group by academic performance, depending on their behavioral patterns, are constructed. Originality. This article proposes the production models of building up student behavior during the test (exam). Practical value. Formalization of student distribution models by typological groups depending on academic performance can be used in the development of an automated tool for creating information technology to form individual methods of teaching students