International Journal of Interactive Mobile Technologies (iJIM) – eISSN: 1865-7923 – Vol  16 No  21 (2022)


Paper—Using Machine Learning to Analyze the Learning Process for Solving Mathematical Problems 

Using Machine Learning to Analyze the Learning Process 
for Solving Mathematical Problems  

https://doi.org/10.3991/ijim.v16i21.36065  

Nauryzbayev Bauyrzhan1(), Baygamitova Sakysh2, Akhmetova Zhanar1,3,  
Pak Nikolay3,4, Karipzhanova Ardak1,5, Urazbaeva Kumys1,6 

1 Kazakh Humanitarian Law Innovative University, Semey, Kazakhstan  
2 Shakarim University, Semey, Kazakhstan  

3 Krasnoyarsk State Pedagogical University named after V.P. Astafyev, Krasnoyarsk, 
 Russian Federation 

NBAcom_1989@mail.ru  

Abstract—The relevance of the research under consideration is due to the 
need to improve the efficiency of the analysis of the quality, and completeness of 
the knowledge obtained by students when solving computational problems, the 
example problems in mathematics. The theoretical argumentation is proposed 
and the practical implementation of an intelligent automated analytical system 
for analyzing the quality and forecasting the content of educational material and 
the trajectories of the student's learning direction is described. The relevance of 
the research is the creation of neural network algorithms that allow analyzing the 
dynamics of changes in the student's level of formation of skills to solve 
arithmetic problems. The methods of analyzing the assimilation of educational 
information and methods of personalized construction of the curriculum for each 
student are substantiated. 

Keywords—artificial intelligence, informatics, mental schemas, pedagogy  

1 Introduction 

From research, it is known that robots take on thousands of routine operations and 
can displace many low-skilled jobs in developed and developing countries. At the same 
time, advanced technologies open up new opportunities, creating conditions for the 
emergence of new and transformed jobs, increasing productivity, and improving the 
efficiency of public services [1-4]. 

Studies have repeatedly proved that the quality of teaching and the professionalism 
of teachers are the most important school factors determining the academic success of 
students [5,6]. It is the teacher who is the key factor in the process of developing and 
translating knowledge, skills, and abilities to students. As noted by the British scientist 
Lawrence Stenhouse, one of the founders of the direction of studying the practice of 
teachers, "in the end, it is teachers who understand the school world from the inside 
who will be able to change it". The difference in training between a more capable and 
a less strong teacher for a student can be an entire academic year. The use of automated 

114 http://www.i-jim.org

https://doi.org/10.3991/ijim.v16i21.36065
mailto:NBAcom_1989@mail.ru


Paper—Using Machine Learning to Analyze the Learning Process for Solving Mathematical Problems 

intelligent analytical software by teachers in educational activities will allow improving 
the quality of teaching.  

That is why the article emphasizes the fundamental role of an intellectual program 
in analyzing the learning process for solving mathematical problems – a problem that, 
by its very nature, does not allow simple and perceptive solutions. An analysis of 
international experience shows that the success of the education system is guaranteed 
if teachers use modern intellectual teaching tools [7]. In the era of digital learning, 
teachers have less and less face-to-face contact with students. This will increasingly 
happen on digital platforms in the form of online training. This will provide many 
problems for analyzing the quality of learning material assimilation. 

Important principles of education are based on freedom, integrity, and consistency. 
In particular, it is necessary to observe the consistency concerning growing up [8]. For 
the most effective educational program, the teacher should devote more time to 
professional development, and routine teaching tasks such as criterion assessment and 
analysis of student performance should be done by intelligent software — within the 
framework of a specific curriculum. As the complexity of the software becomes more 
complex, the teacher becomes less dependent on the control of the student's knowledge, 
remaining within the curriculum, and will be able to free up time for a creative approach 
to the lesson. At the same time, intelligent software for the analysis and control of 
student performance bears a great responsibility for the quality of the analysis of student 
performance. But, as always, the principle applies here: the level of freedom cannot 
increase without a parallel increase in responsibility. These two concepts in 
consciousness should not be separated from each other. 

Forecasting and improving the quality of the results of students ' educational 
activities is one of the main components of the educational process. Currently, digital 
learning resources are becoming intellectual, and methods of teaching knowledge that 
ensure the effectiveness of learning require rethinking [9, 10]. The covid-19 pandemic 
accelerated this process and proved that the existing expert systems are not yet 
sufficiently refined for their application during distance education. The most important 
component of the methods of effective teaching of students in natural science 
disciplines is the" step-by-step " method of learning to solve computational problems.  

Modern trends in education and society indicate the demand for intelligent 
educational process management systems as a tool for analyzing the educational 
process and controlling the quality of acquired knowledge [11]. In this regard, the 
problem of the formation and development of students ' skills and abilities to solve 
computational problems in natural science areas becomes extremely relevant. Despite 
the large number of methods of teaching problem solving with different methodological 
approaches, the issues of formalization and personification of the process of organizing 
intellectual analysis of students ' progress in solving problems are insufficiently studied 
and are of considerable interest from the standpoint of intelligent automation of such 
activities. Indeed, in most cases, the effectiveness of known teaching methods depends 
on the skill and experience of the teacher, these methods are time-consuming and 
require a lot of time, especially in terms of diagnosing learning outcomes [12].  

In this regard, the creation of computer training, and diagnostic systems that provide 
the definition, analysis, and assessment of the formation of the ability to solve 

iJIM ‒ Vol. 16, No. 21, 2022 115



Paper—Using Machine Learning to Analyze the Learning Process for Solving Mathematical Problems 

computational problems is extremely relevant for the theory and practice of digital 
learning. 

Despite the significant technological and didactic potential of modern information 
and communication technologies, their use in training students to solve problems is 
unsatisfactory for several reasons [13-15]: 

• Firstly, Digital means of forming the ability to solve problems are traditionally more 
instructive, reference in nature, and controlling means, as a rule, assess the level of 
formation of this skill according to the final result, according to the "black box" 
model. 

• Secondly, the existing automated training systems are poorly connected with the 
principles of personality-oriented, personalized learning. 

• Thirdly, the motivation and effectiveness of students ' independent work on solving 
computational problems significantly depend on flexible, adaptive, and unobtrusive 
external management, as well as on convenient and visualized self-management of 
educational activities according to the "white box" model. 

• Fourth, the teacher makes a physical analysis of the progress of each student based 
on their qualifications. As a result, as students grow and deepen into the educational 
material, the teacher loses the objectivity of the student's progress in mastering the 
educational material. 

The purpose of this article is the theoretical substantiation of the model of 
personalized work of students on the formation of the ability to solve computational 
problems from the standpoint of a mental approach and the practical implementation of 
an automated training and diagnostic system with a visualized mechanism of self-
management of educational activities on a mirror subject-object-subject interaction. 

1.1 Review of the scientific literature on the problem 

With the development of digital technologies and their application in the life of 
society, the problem of assimilation of knowledge in solving computational problems 
in the natural science field is becoming more acute. Problem-solving is a highly 
individualized skill, which largely depends on abstract thinking and previously learned 
educational materials [16]. 

The analytical work of the teacher in the educational activities of students plays a 
significant role in the learning process. Its effectiveness largely depends on the didactic 
qualities of digital learning tools. As a rule, most of these educational resources are 
aimed at teaching according to the principles of modern didactics, forming and 
developing the required abilities and competencies [17]. 

Recently, researchers are increasingly paying attention to innovative methods and 
means of teaching in the context of digital learning (e-learning) [18]. However, as real 
practice shows, digital learning is not always successfully and effectively implemented 
in the educational process [19]. Many teachers have high hopes for the personification 
of learning from the perspective of a student-centered paradigm of education [20-24]. 
Also, teachers are in urgent need of intelligent analytical assistants who could facilitate 
their work and allow them to devote more time to the learning process itself.  

116 http://www.i-jim.org



Paper—Using Machine Learning to Analyze the Learning Process for Solving Mathematical Problems 

The quality of the student's assimilation of educational material for solving problems 
depends on several factors, including the following: 

• student independence is expressed in the responsibility of students, not teachers, for 
educational results; 

• personification of the educational process — the teacher models the individual 
educational trajectory of each of his students based on his capabilities; 

• orientation to significant achievements in independent learning is a psychological 
and pedagogical phenomenon that contributes to the manifestation of interest and 
internal motivation; 

• self-diagnosis and comparison of the achieved learning results with reference ones 
is also a psychological and pedagogical phenomenon that enhances self-expression 
and the importance of one's personality. 

The effectiveness of training students in the educational process depends not only 
on the ability of teachers to competently manage the training of students and on the 
degree of preparation of students for this type of activity but also to a greater extent on 
the appropriate means of teaching [25]. 

By transferring pedagogical functions to oneself, a person thereby masters the 
system of appropriate "meta-cognitive skills" [26]. It is more convenient to provide 
conditions for the organization of personalized learning based on a cognitive or mental 
approach. From a cybernetic point of view, the student can be considered a "black box" 
model. The teacher exerts training influences on him, trying to form the required 
properties (learning outcomes). It is advisable to carry out coordinated training with the 
help of feedback control. Traditionally, the control is implemented by analyzing the 
protocol of observations of the "black box". At the same time, the protocol is 
understood as a list of effects on the "black box" and the corresponding reactions. Well-
known methods of learning, including innovative e-learning methods, are based on the 
analysis of the protocol (Figure 1a). The cognitive approach allows us to model the 
mental structure of thinking, and to study the issues of its development, which can 
provide the opportunity to build an educational process according to the "white box" 
model. In this case, the learning process is reduced to the formation of the required 
mental structures in the student (Figure 1b). 

 
Fig. 1a. Training on the "black box" model  

with the analysis of the protocol 
Fig. 1b. Training on the "white box" 

model 
*Designations: O-training, Y-student, OS-feedback, K-control, P-protocol, BYA-white box 

iJIM ‒ Vol. 16, No. 21, 2022 117



Paper—Using Machine Learning to Analyze the Learning Process for Solving Mathematical Problems 

The central concept in the mental approach to learning is the concept of "mental 
schema", introduced by W. By Neisser[13]. At the physical level, the mental circuit is 
a set of structural elements and processes in the nervous system that control thinking 
and activity. There are mental schemes that are formed when solving computational 
problems and managing this activity. The purpose of training in this case is the 
purposeful formation of mental schemes responsible for solving computational 
problems. At the conceptual level, the mental schema model is represented as a graph 
containing terminal (data objects and goals) and non-terminal vertices (a lower-level 
mental schema) [27]. 

2 Methodology 

Digital learning tools of the new generation should carry the functions not only of 
presenting educational information but also of developing mental operations and their 
analysis. They allow the student to independently manage the learning process. They 
should provide different options for adjusting the text to the psychological preferences 
of the student. Due to the hierarchy of cognitive qualities of a person, it is important to 
provide a graph structure and collapsibility of certain text fragments that are of a 
clarifying nature in the interface of a digital publication. Such a view will allow you to 
simulate different training routes for mastering a given topic. At the same time, the 
student can change and adjust the route of study in the course of educational activities, 
depending on his motivation, acquired experience, and claims for the result of training 
[28]. 

The main advantage of digital cognitive learning tools is interactivity and 
visualization. All sessions of working with trainees should be remembered in a special 
database with a statistical mechanism. This is necessary for the subsequent stages of 
training, in particular, it will be possible to generate more often those tasks in which 
most users had difficulties, and where they made more mistakes. It is preferable to use 
a mental approach to build adaptive and independent teaching of students and 
schoolchildren to solve problems using the "white box" model [29]. The experience of 
developing and using training tools using mental schemes has shown their high 
efficiency in the educational process [30]. 

In our model, the initial data set contains linear algebra data from the 7th-grade 
program, as well as the name of topics and scientific problems for each row. For a more 
accurate analysis, it was decided to use the Pandas library. To get the right result, 
converting training data to the right level of detail for analysis or a machine learning 
project is a common task and often requires some level of pre-processing and 
experimentation. 

One of the difficult aspects of designing natural language processing is that all the 
analyzed texts will contain a lot of words and elements that do not provide any 
meaningful information in terms of identifying the underlying structure or topic. These 
can be general words, such as "I" or "my", as well as specific words that are often found 
in the entire collection of the text being studied. Another very difficult problem is the 
recognition of mathematical formulas written in the MathType editor. 

118 http://www.i-jim.org



Paper—Using Machine Learning to Analyze the Learning Process for Solving Mathematical Problems 

2.1 Making a model 

To solve the problem of binary classification of objects, it is proposed to use various 
machine learning methods, as well as their combinations (an aggregated classifier). It 
is impractical to talk about the effectiveness of any of the methods considered, since 
we can get different results for different samples, even for different parts of the same 
sample. You can only introduce several restrictions on the selection related to the 
peculiarity of a particular method or field of research, for example, the logistic model 
is sensitive to correlation between factors, so the presence of strongly correlated input 
variables is unacceptable in the model; the support vector machine is sensitive to noise 
and data standardization; to apply the Bayesian approach, it is necessary first to bring 
the initial data to an interval scale, so that the variables are discrete, otherwise, this may 
lead to the loss of significant information, etc. As a result, the developed program will 
select the best (in the sense of the specified criteria) combination of the methods 
involved, that is, the optimal aggregated classifier [31]. 

To analyze the curriculum, which in a certain period will analyze the tasks solved 
by the student, a thematic model was created based on the curriculum for the 7th grade. 
"Thematic modeling" is the process by which we find hidden topics in a series of 
documents. This means that according to the data presented from several documents—
in this case, the mathematics curriculum for the 7th grade — we find sets of words, and 
formulas that often coexist, forming coherent "topics". After determining these topics, 
formulas, and words, we trace how common these topics are in the student's work over 
time, indicating the progress of the student's academic performance in a given 
curriculum. 

The first step to building a topic model is to extract objects from the corpus for 
modeling. In the process of natural language processing, the preferred technique is the 
"bag of words" model. The "bag of words" model summarizes the frequencies of each 
word in a document, with each unique word being its characteristic, and its frequency 
being the value [32]. 

An even better feature extraction tool that does a deep analysis is the term 
"frequency-inverse frequency" of the document. This method shows the number of 
words that appear very often in the corpus, since they should be less likely to give an 
idea of specific topics of the document, given how common this word is. Using its tf-
idf, we can determine the features for each element of the training material, which can 
represent how important each word is for our analysis program. 

 
Fig. 2. The "frequency-inverse frequency" method of the document 

After we have defined the methods for analyzing the text in the curriculum, we create 
a thematic model! The modeling method that was used for this project is NMF. NMF, 
or non-negative matrix factorization. We will use this algorithm to identify our topics 

iJIM ‒ Vol. 16, No. 21, 2022 119



Paper—Using Machine Learning to Analyze the Learning Process for Solving Mathematical Problems 

or groups of words and formulas that occur, as well as to determine the prevalence of 
these topics in each student's decision. 

The first part of the NMF output is the text. At this stage of the machine learning 
process, we get the opportunity to use the data of the training program for analysis. 
After checking the text in each topic, you can make a responsible decision about which 
topic or idea is represented by words and we can post a visual analysis of the learned 
material. 

Another part of the NMF output is the "Document-topic" matrix. In this matrix, each 
row is a topic of a section, each column is a task, and the value is a relative assessment 
of how much this topic exists in a particular student's work and how it was completed. 

In the course of the work, the task was to see how often the materials of a certain 
section are used in each student's work, and it was also necessary to identify the criteria 
assessment and identify the most successful descriptors for each topic. During the 
analysis of the algorithms used with the threshold for evaluating the student's work, it 
was decided to set the threshold at the level of 0.1. After the training, we had a 
transformed matrix of the topics of the sections and their corresponding tasks ready. At 
the next stage, it was necessary to train the program to effectively evaluate and 
determine the descriptors in the database.  

3 Results 

To develop a mental scheme of skills to solve computational physical problems, we 
will highlight elementary phenomena and processes. We will describe each 
phenomenon by an appropriate physical model, which is generally characterized by 
several quantities. They, in turn, will be interconnected by some mathematical law, 
formula, or equation (mathematical model). Let's call the combination of physical and 
mathematical models a computational primitive. 

After building and teaching a complete thematic model, for a more accurate analysis 
of problem solutions by the student, it was necessary to create a new model that would 
consider deeper relationships between individual objects, rather than general topics of 
the curriculum. To do this, a modeling method called "word2vec" was used. With this 
model, we can match each object that appears in the texts of the curriculum. As a result 
of the program, you can look at the similarity of objects. Such a mapping of a word into 
a vector space is called an embedding of a word. With the help of this model, we can 
see how similar some words are to each other in terms of the style of writing problem 
solving by the student and the model embedded in the knowledge base. 

By observing the words and objects that are grouped in the tSNE below, we get an 
idea of how well the problem was solved and reveal how complete the student received 
knowledge on a given topic [33, 34]. It should be remembered that the words are 
grouped based on the similarity of their attachments in the knowledge base of the 
intellectual program for analyzing the correctness of the tasks being solved by the 
student. 

120 http://www.i-jim.org



Paper—Using Machine Learning to Analyze the Learning Process for Solving Mathematical Problems 

4 Discussion 

The possibilities of natural language processing are endless, and the idea is that it 
can give us a deeper understanding of how it will be possible to improve educational 
activities in the future [20]. The proposed method of creating an intelligent automated 
training and diagnostic system for learning how to solve computational problems has 
several advantages over existing analogs. 

Firstly, the system ensures the implementation of the principles of personalized 
learning and has a high degree of adaptability for each individual user. Secondly, the 
system motivates the effectiveness of teaching students to solve computational 
problems due to flexible and unobtrusive external control, as well as due to a convenient 
and visualized mechanism for analyzing the development of educational material 
according to the "white box" model [15]. Thirdly, the system provides for the forgetting 
factor and assumes its repeated use to achieve the planned learning result by the student. 
Tasks are generated based on statistical data accumulated during the work of all users 
of the system. 

5 Conclusion 

The proposed method can be used to create structural-mental schemes not only for 
sections of the mathematics course but also for other disciplines in which solving 
computational problems is a common activity (chemistry, physics, computer science, 
etc.). The representation of structural-mental schemes in the form of a graph is very 
convenient for software implementation. The level of formation of the ability to solve 
problems using structural-mental schemes is determined by the superposition of 
particular structural-mental schemes of individual tasks that the student has solved. In 
the case when private schemes completely cover the general structural and mental 
scheme, we can talk about the formation of the ability to solve computational problems 
in the subject area represented by this scheme. 

The rating, unlike the methods for determining latent characteristics developed in 
IRT, allows you to determine the difficulty of tasks and the readiness of the student 
dynamically, in the learning process. Its use does not imply the presence of a large bank 
of calibrated tasks and estimates of various parameters that appear in other models. This 
allows you to organize an adaptive selection of tasks quickly, simply, and cheaply. 

An intelligent analytical diagnostic system for teaching elementary mathematics 
problem-solving was implemented under the Windows operating system. It will be 
posted on the Internet after full approbation and receipt of the author's certificate. The 
analysis of the system will allow us to further modernize the methodology for creating 
cognitive tools for analyzing the learning process and achieve more effective results of 
mastering the educational material by students for the formation of skills to solve 
computational problems. 

iJIM ‒ Vol. 16, No. 21, 2022 121



Paper—Using Machine Learning to Analyze the Learning Process for Solving Mathematical Problems 

6 References 

[1] World Bank. (2021). Changing the Nature of Work. Accessed: Apr. 16, 2021. [Online]. 
Available: https://projects.vsemirnyjbank.org/ru/projects-operations/projects-list?os=0 

[2] Akcil, U., Uzunboylu, H., & Kinik, E. (2021). Integration of Technology to Learning-
Teaching Processes and Google Workspace Tools: A Literature Review. Sustainability, 
13(9), 5018. https://doi.org/10.3390/su13095018  

[3] Pascu, L., Simo, A., & Vernica, A. M. (2019). Integrating Microsoft IoT, machine learning 
in a large-scale power meter reading. International Journal of New Trends in Social 
Sciences, 3(1), 10–16. https://doi.org/10.18844/ijntss.v3i1.3815  

[4] Bhuyan, M. H., & Tamir, A. (2020). Evaluating COs of computer programming course for 
OBE-based BSc in EEE program. International Journal of Learning and Teaching, 12(2), 
86–99. https://doi.org/10.18844/ijlt.v12i2.4576  

[5] Seree, M. A. A., Mohamed, M. I., & Mustafa, M. E. D. (2021). Progressive computer test 
capabilities of the candidates proposed for the post of physical education teacher. 
Contemporary Educational Researches Journal, 11(1), 09–17. https://doi.org/10.18844/ 
cerj.v11i1.4729  

[6] Salama, R., Elsayed, M., & Shadi, M. A. (2021). Learning programming languages by using 
e-learning technology. Global Journal of Computer Sciences: Theory and Research, 11(2), 
100–108. https://doi.org/10.18844/gjcs.v11i2.5289  

[7] Xu, W. (2021). Computer forensics teaching for undergraduate students. Global Journal of 
Information Technology: Emerging Technologies, 11(2), 29–34. https://doi.org/10.18844/ 
gjit.v11i2.5266  

[8] Issayev, G., Bagdat, B., Abay, D., Aizhan, S., Baimanova, L., & Almazhai, Y. (2022). 
Application of information technologies in distance learning in the field of higher education. 
World Journal on Educational Technology: Current Issues, 14(4), 1017–1024. 
https://doi.org/10.18844/wjet.v14i4.7650  

[9] Benmammar, S. (2020). Teaching English for specific purposes to computer science 
students with reading difficulties. Global Journal of Foreign Language Teaching, 10(3), 
159–166. https://doi.org/10.18844/gjflt.v10i3.5072  

[10] Goceri, E. (2018). Advances in digital pathology. International Journal of Emerging Trends 
in Health Sciences, 1(2), 33–39. https://doi.org/10.18844/ijeths.v1i2.3107  

[11] Pak, N. I., & Asaulenko, E. V. (2018). Personification of students' independent work on the 
formation of the ability to solve computational problems on the basis of an automated 
teaching and diagnostic system. Computer Science and Education, (8), 26-32. 
https://doi.org/10.32517/0234-0453-2018-33-8-26-32  

[12] Nurzhanov, C., Pidlisnyuk, V., Naizabayeva, L., & Satymbekov, M. (2021). Research and 
trends in computer science and educational technology during 2016-2020: Results of a 
content analysis. World Journal on Educational Technology: Current Issues, 13(1), 115–
128. https://doi.org/10.18844/wjet.v13i1.5421  

[13] Moghadamizad, Z., Mowlaie, B., & Rahimi, A. (2020). An inquiry on publishers’ criteria 
for recruitment of translators. International Journal of New Trends in Social Sciences, 4(2), 
77–93. https://doi.org/10.18844/ijntss.v4i2.5127  

[14] Bhuyan, M. H., & Tamir, A. (2020). Evaluating COs of computer programming course for 
OBE-based BSc in EEE program. International Journal of Learning and Teaching, 12(2), 
86–99. https://doi.org/10.18844/ijlt.v12i2.4576  

[15] Salama, R., Qazi, A., & Elsayed, M. (2018). Online programming language—Learning 
management system. Global Journal of Information Technology: Emerging Technologies, 
8(3), 114–123. https://doi.org/10.18844/gjit.v8i3.4051  

122 http://www.i-jim.org

https://projects.vsemirnyjbank.org/ru/projects-operations/projects-list?os=0
https://doi.org/10.3390/su13095018
https://doi.org/10.18844/ijntss.v3i1.3815
https://doi.org/10.18844/ijlt.v12i2.4576
https://doi.org/10.18844/cerj.v11i1.4729
https://doi.org/10.18844/cerj.v11i1.4729
https://doi.org/10.18844/gjcs.v11i2.5289
https://doi.org/10.18844/gjit.v11i2.5266
https://doi.org/10.18844/gjit.v11i2.5266
https://doi.org/10.18844/wjet.v14i4.7650
https://doi.org/10.18844/gjflt.v10i3.5072
https://doi.org/10.18844/ijeths.v1i2.3107
https://doi.org/10.32517/0234-0453-2018-33-8-26-32
https://doi.org/10.18844/wjet.v13i1.5421
https://doi.org/10.18844/ijntss.v4i2.5127
https://doi.org/10.18844/ijlt.v12i2.4576
https://doi.org/10.18844/gjit.v8i3.4051


Paper—Using Machine Learning to Analyze the Learning Process for Solving Mathematical Problems 

[16] Akula, S. R. (2021). Semi-supervised machine learning approach for DDOS detection. 
International Journal of Innovative Research in Education, 8(1), 27–35. https://doi.org/ 
10.18844/ijire.v8i1.6445  

[17] Pogorelskaya, I. L. S. I. A. R., & Várallyai, L. Á. S. Z. L. Ó. (2020). Trends In Education 
4.0. THE ANNALS OF THE UNIVERSITY OF ORADEA, 29(2020), 367.  

[18] Bazhenova, I., Babich, N., & Pak, N. (2022). From projective-recursive learning technology 
to mental didactics. liters. 

[19] Guri-Rosenblit, S., & Gros, B. (2011). E-learning: Confusing terminology, research gaps, 
and inherent challenges. International Journal of E-Learning & Distance Education/Revue 
Internationale du e-learning et la formation à distance, 25(1). 

[20] Pack, N. I., Petrova, I. A., Pushkaryeva, T. P., & Hegai, L. B. (2018). Organization of 
Student-Centered Learning for Students on the Basis of Transformable Academic Course. 
Astra Salvensis. 

[21] Trinidad, J. E. (2020). Understanding student-centred learning in higher education: students’ 
and teachers’ perceptions, challenges, and cognitive gaps. Journal of Further and Higher 
Education, 44(8), 1013-1023. https://doi.org/10.1080/0309877X.2019.1636214  

[22] Hannafin, M. J., & Hannafin, K. M. (2010). Cognition and student-centered, web-based 
learning: Issues and implications for research and theory. In Learning and instruction in the 
digital age (pp. 11-23). Springer, Boston, MA. https://doi.org/10.1007/978-1-4419-1551-
1_2  

[23] Crumly, C., D'Angelo, S., & Dietz, P. (2014). Pedagogies for student-centered learning: 
Online and on-ground. Augsburg Fortress Publishers. https://doi.org/10.2307/j.ctt9m0skc  

[24] Zakharova, I. G., Lapchik, M. P., Pak, N. I., Ragulina, M. I., Timkin, S. L., Udalov, S. R., 
... & Henner, E. K. (2017). Modern problems of informatization of education. 

[25] Ryabov G.P. (2009). New Information Technologies for the New Teacher, New Information 
Technologies, vol. 4, pp. 30-35, 2009, Accessed: Jul. 06, 2021. Available: https://elibrary.ru/ 
download/elibrary_14751304_34849374.pdf 

[26] Mushtavinskaya, I. V. (2011). The use of reflective technologies in the development of 
students' ability to self-education as a pedagogical problem. In Pedagogy: tradition and 
innovation (pp. 146-151). 

[27] Neisser, W. (1981). Knowledge and reality. Meaning and principles of cognitive 
psychology. Moscow: Progress, 230, 5. 

[28] Walz, J. A. A Guide to the Future, PMLA/Publications of the Modern Language Association 
of America, vol. 56, no. S1, pp. 1324–1334, 1941. https://doi.org/10.2307/458906  

[29] Asaulenko, E. V. (2017). Formation of the student's abilities to solve computational physical 
problems based on mental schemes. Educational Informatics, (2), 11-19. 

[30] Bazhenova, I. V. (2015). Features of programming teaching methodology based on 
projective-recursive strategy and cognitive technologies. Pedagogical Education in Russia, 
(3), 52-57. 

[31] Alekseeva, V. A. (2015). Using machine learning methods in binary classification problems. 
Automation of control processes, (3), 58-63. 

[32] McTear, M., Callejas, Z., & Griol, D. (2016). Spoken language understanding. In The 
Conversational Interface (pp. 161-185). Springer, Cham. https://doi.org/10.1007/978-3-319-
32967-3_8  

[33] Lee, D. D., & Seung, H. S. (1999). Learning the parts of objects by non-negative matrix 
factorization. Nature, 401(6755), 788-791. https://doi.org/10.1038/44565  

[34] Aonishi, T., Maruyama, R., Ito, T., Miyakawa, H., Murayama, M., & Ota, K. (2022). 
Imaging data analysis using non-negative matrix factorization. Neuroscience Research, 179, 
51-56. https://doi.org/10.1016/j.neures.2021.12.001  

iJIM ‒ Vol. 16, No. 21, 2022 123

https://doi.org/10.18844/ijire.v8i1.6445
https://doi.org/10.18844/ijire.v8i1.6445
https://doi.org/10.1080/0309877X.2019.1636214
https://doi.org/10.1007/978-1-4419-1551-1_2
https://doi.org/10.1007/978-1-4419-1551-1_2
https://doi.org/10.2307/j.ctt9m0skc
https://elibrary.ru/download/elibrary_14751304_34849374.pdf
https://elibrary.ru/download/elibrary_14751304_34849374.pdf
https://doi.org/10.2307/458906
https://doi.org/10.1007/978-3-319-32967-3_8
https://doi.org/10.1007/978-3-319-32967-3_8
https://doi.org/10.1038/44565
https://doi.org/10.1016/j.neures.2021.12.001


Paper—Using Machine Learning to Analyze the Learning Process for Solving Mathematical Problems 

[35] Kurzweil R. and Kurzweil R. Ėvoli︠ u︡t︠ s︡ii︠ a︡ razuma, ili Beskonechnye vozmozhnosti 
chelovecheskogo mozga, osnovannye na raspoznavanii obrazov. 

7 Authors 

Nauryzbayev Bauyrzhan is a Ph.D. candidate, specialty "Informatics", Kazakh 
Humanitarian Law Innovative University, Abay street 107, 071405, Semey city, 
Kazakhstan (NBAcom_1989@mail.ru, https://orcid.org/0000-0003-4935-5972). 

Baygamitova Sakysh is a 2nd-year master's student, specialty "Training of 
mathematics teachers", Shakarim University, Abay street 107, 071405, Semey city, 
Kazakhstan. (baygamitova90@mail.ru, https://orcid.org/0000-0002-7184-808X). 

Akhmetova Zhanar, Ph.D., is an acting Associate Professor of the Department 
"Information Security" L. N. Gumilyov Eurasian National University, Glinka street, 20 
"a", 071412, Semey city, Kazakhstan (Akhmetova_zhzh@enu.kz, https://orcid.org/ 
0000-0002-5483-5260). 

Pak Nikolay, Krasnoyarsk State Pedagogical University named after V.P. Astafyev, 
Ada Lebedeva str., 89, Krasnoyarsk city, 660049, Russian Federation (nik@kspu.ru, 
https://orcid.org/0000-0003-2105-8861).  

Karipzhanova Ardak, Ph.D. is a Dean of the Faculty of Information Technology 
and Economics, Senior Lecturer of the Department of Information and Technical 
Sciences, Kazakh Humanitarian Law Innovative University, Abay street 107, 071405, 
Semey city, Kazakhstan (kamilakz2001@mail.ru, ORCID - https://orcid.org/0000-
0002-0113-6132).  

Urazbaeva Kumys is a candidate of Physical and Mathematical Sciences, Associate 
Professor, at Kazakh Humanitarian Law Innovative University, Abay street 107, 
071405, Semey city, Kazakhstan. (urazbaeva57@mail.ru, ORCID - https://orcid.org/ 
0000-0002-8296-8394). 

Article submitted 2022-09-10. Resubmitted 2022-10-14. Final acceptance 2022-10-14. Final version 
published as submitted by the authors. 

124 http://www.i-jim.org

mailto:NBAcom_1989@mail.ru
https://orcid.org/0000-0003-4935-5972
mailto:baygamitova90@mail.ru
https://orcid.org/0000-0002-7184-808X
mailto:Akhmetova_zhzh@enu.kz
https://orcid.org/0000-0002-5483-5260
https://orcid.org/0000-0002-5483-5260
mailto:nik@kspu.ru
https://orcid.org/0000-0003-2105-8861
mailto:kamilakz2001@mail.ru
https://orcid.org/0000-0002-0113-6132
https://orcid.org/0000-0002-0113-6132
mailto:urazbaeva57@mail.ru
https://orcid.org/0000-0002-8296-8394
https://orcid.org/0000-0002-8296-8394