S.S. Kitov1, I.A. Prokopenkov2, S. Gribchenko3, V.V. Kulikov4, O.V. Platonova5

1,2 Military Academy of Military Air Defense of the Armed Forces of the Russian Federation (Smolensk, Russia)
3 Military Institute of Physical Culture (St. Petersburg, Russia)
4 Joint Stock Company «Concern «Morinformsystem – Agat» (Moscow, Russia)
5 MIREA – Russian Technological University (Moscow, Russia)
1 kot_vifk@mail.ru, 2 prokopenkoff.ivan@yandex.ru, 4 vlk6161@yandex.ru

The program proposed in the article for forecasting the optimal variant of a set of educational tools has a high generality. The principles laid down in the program make it possible to apply it in solving problems of evaluating organizational and technical systems, in substantiating the appearance of educational and training complexes, etc. As an illustrative example, the use of a forecasting program as an effective educational tool for a specific group of students is considered. Modern research records the positive role of physical culture in the formation of students' values such as teamwork, cooperation, mutual support and the ability to achieve a common goal. However, selecting the optimal option for a comprehensive group exercise, taking into account the physical and psychological characteristics of a particular study group, is a multi-criteria classification task that requires significant time investment by the teacher. The article presents the author's prediction program based on the machine learning model "Gradient boosting", which allows to determine the best option for a set of group exercises to increase the cohesion of the cadet team. The program uses average baseline data on indicators of cohesion (competition, adaptation, compromise, avoidance, cooperation) and indicators of physical fitness (running 60 m, pull-up on the crossbar, running 1000 m). After selecting the module of the curriculum (athletics, sports, swimming, gymnastics) and starting the calculation, the trained ensemble model is activated. The validity of the choice of gradient boosting is ensured by its ability to construct complex nonlinear separation surfaces and a built-in regularization mechanism that reduces the risk of overfitting. The advantage of the program is the use of empirical data obtained during pedagogical experiments that have proven the effectiveness of the proposed complexes. The practical significance lies in reducing the teacher's preparation time by 8-10 times, while ensuring that the educational goal of team building is achieved.

Kitov S.S., Prokopenkov I.A., Gribchenko S., Kulikov V.V., Platonova O.V. A program for predicting the optimal variant of a set of educational tools based on machine learning and artificial intelligence // Science Intensive Technologies. 2026. V. 27. № 4.
P. 89−95. DOI: https://doi.org/ 10.18127/j19998465-202604-09

1. Kozlov A.P., Popov V.A. Vliyanie fizicheskoj podgotovki voennosluzhashchih na boevuyu splochennost' voinskih kollektivov. Simvol nauki. 2020. № 2. S. 101–104 (In Russian).
2. Ostapenko R.I., Nebol'sin A.A. Issledovanie splochennosti voinskoj gruppy s pomoshch'yu matematicheskih metodov. Psihologiya, sociologiya i pedagogika. 2014. № 4. URL: https://psychology.snauka.ru (data obrashcheniya: 20.02.2026) (In Russian).
3. Bahtin A.A., Savel'ev V.Yu. Primenenie metodov mashinnogo obucheniya dlya prognozirovaniya uspeshnosti uchebnoj deyatel'nosti. Informatika i obrazovanie. 2023. № 2. S. 45–54 (In Russian).
4. Hajkin S. Nejronnye seti: polnyj kurs. Izd. 2-e. M.: Vil'yams. 2021. 1104 s. (In Russian).
5. Chen T., Guestrin C. XGBoost: A Scalable Tree Boosting System. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. San Francisco, 2016. P. 785–794.