Identification of Student Learning Styles Using the Dempster-Shafer Theory Algorithm

Abstract: In the school environment, learning is the interaction between educators and students that is done consciously, planned both inside and outside the room to improve the ability of students. But a person's ability to learn, understand and absorb lessons is different in level. Some are fast, moderate and some are very slow. Therefore, they often have to take different ways to understand the same information or lessons. This study aims to identify student learning styles by implementing the Dempster-Shafer Theory algorithm to help teachers determine the learning model that is suitable for each student. Students who were the object of this study were students of SMA N 1 Klampok Purwareja. The Dempster-Shafer Theory method can be used for the representation, combination and propagation of uncertainty, characteristics that are institutively following the way of thinking of an expert, this method can also distinguish the uncertainty and ignorance of the expert. So that research can be directed, the researcher uses the system development method ESDLC (Expert System Development Life Cycle) with stages of Assessment, Knowledge Acquisition, Design, Testing, Documentation, and Maintenance. The results of this study are student learning style learning applications that make it easy for teachers to determine learning models that are suitable for student. Abstrak: Di lingkungan sekolah, belajar adalah interaksi antara pendidik dengan peserta didik yang dilakukan secara sadar, terencana baik didalam maupun di luar ruangan untuk meningkatkan kemampuan peserta didik. Namun kemampuan seseorang dalam belajar, memahami dan menyerap pelajaran berbeda tingkatnya. Ada yang cepat, sedang dan ada pula yang sangat lambat. Karenanya, mereka seringkali harus menempuh cara berbeda untuk bisa memahami sebuah informasi atau pelajaran yang sama. Penelitian ini bertujuan untuk mengidentifikasi gaya belajar siswa dengan mengimplementasikan algoritma Dempster-Shafer Theory agar dapat membantu guru menentukan model pembelajaran yang cocok untuk masing-masing siswa. Siswa yang dijadikan objek penelitian ini adalah siswa SMA N 1 Klampok purwareja. Metode Dempster-Shafer Theory dapat digunakan untuk representasi, kombinasi dan propogasi ketidakpastian, karakteristik yang secara instutitif sesuai dengan cara berfikir seorang pakar, metode ini juga dapat membedakan ketidakpastian dan ketidaktahuan dari pakar. Agar penelitian dapat terarah maka peneliti menggunakan metode pengembangan sistem ESDLC (Expert System Development Life Cycle) dengan tahapan Assesment, Knowledge Acquisition, Design, Testing, Documentation, Dan Maintenance. Hasil penelitian ini sebuah aplikasi identifikasi gaya belajar siswa sehingga mempermudah guru untuk menentukan model pembelajaran yang cocok bagi siswa.


Introduction
Learning is a change in behaviour or appearance, with a series of activities, for example, by learning to read, observe, listen, imitate, and so forth. Learning will also be better if the subject of learning experiences or does it [1]. Everyone has a curiosity about something. These feelings and needs drive people always to learn. So, learning is an effort of people to know something that is not yet known, and everyone does it.
In the school environment, learning is the interaction between educators and students that is done consciously, planned both inside and outside the room to improve the ability of students. Learning at school means that the interaction between the teacher and students is carried out consciously and planned, which is carried out both in the classroom and outside the classroom to improve student abilities [2].
But a person's ability to learn, understand and absorb lessons is different in level. Some are fast, moderate, and some are very slow. Therefore, they often have to take different ways to understand the same information or lessons. According to [3], student success in learning is influenced by external factors and internal factors. Internal factors are factors that originate from within students and one of them is learning style. Learning styles are a combination of ways to absorb and process information.
Based on the research [4] to identify student learning styles by building an expert system with a dynamic knowledge base Knowledge is obtained from various sources, including research conducted by experts in their fields and books related to learning styles. The drawing of conclusions in this expert system uses the forward chaining inference method. This expert system will display questions about the characteristics of the perceived learning style, then to get the final result. In the final result, the expert system will display the types of learning style characteristics possessed by students. To solve the problem of uncertainty in identifying student learning styles, the authors chose to use the Dempster-Shafer Theory Method.
Based on the research [5] Applying the Dempster Shafer Method to an expert system to identify Indihome (Indonesian Digital Home) service interruptions at PT Telkom to continue access to Indihome Services so that it can work well The results of this study is an expert system that diagnoses disturbances Indihome service. The system automatically provides diagnosis results by displaying the type of disorder and its solution based on the symptoms of the disorder experienced. This study concludes that an expert system using the Dempster Shafer method to handle data uncertainty when diagnosing Indihome service interruptions is very helpful in overcoming the problem of decreased service quality.
The choice of Dempster-Shafer Theory as a solution to identifying students' learning styles is because they can make representations, combinations and propagation of uncertainty, where this theory has several characteristics that are institutively following the way of thinking of an expert, but a strong mathematical basis. In this theory can also distinguish between uncertainty and ignorance.
The expert system will be built based on the Website because it is not the same as the desktop and android applications that must be installed, the Website with Responsive Design can also adjust the appearance according to the device used. According to Lutfi [6], Responsive Design is a technique used to make website layouts adjust to the appearance of the visitor's device. A responsive website is an approach that provides site design capabilities in adapting to site user behaviour and site environment.
Which is based on screen size, screen orientation and site container. The way it works consists of a combination of frames and flexible appearance and the use of a creative CSS system. So if a user moves from his computer to another device, the site will change to the right resolution.

Method
Research stages are needed as a framework and guide the research process so that the series of research processes can be carried out in a directed, orderly and systematic way.   The process of making an expert system application selection of tourist destinations using forward rulebased reasoning inference techniques is as follows: a. Determine the object of research that will be applied in the expert system, namely SMA N 1 Klampok, Purwareja. b. Collecting and identifying the types of learning styles, criteria for each type of learning style. c. Analyse the types of learning styles that exist by exploring the criteria of each learning style.
Furthermore, a search is carried out to ascertain whether the expert system can accommodate the criteria and types of learning styles by providing the best solution that must be taken or not. d. The selection of experts whose knowledge will be acquired for interviewing and observing their work in the form of practical experience and understanding of the criteria and types of learning styles. e. Make input and output design. Input expert systems in the form of learning style criteria that use the information provided by students, while the output is in the form of learning styles and suitable learning models. f. The expert system architecture is technically the expert system design architecture identifies the learning style divided into three parts, namely interface, application and output. This interface can be used by users in interacting with expert system applications; this section is used to access the required information objects.

a. Dempster-Shafer Theory
The Dempster-Shafer method was first introduced by Dempster, who conducted uncertainty model experiments with range probabilities rather than as single probabilities. Then in 1976 Shafer published Dempster's theory in a book called Mathematical Theory of Evident. Dempster-Shafer Theory of Evidence shows a way to give the weight of confidence according to the facts collected. In this theory can distinguish between uncertainty and ignorance. Dempster-Shafer theory is a representation, combination and uncertainty propagation, where this theory has several characteristics that are institutively following the way of thinking of an expert, but a strong mathematical basis [9]. In general, the Dempster-Shafer theory is written at an interval: (Pls) will reduce the level of certainty of the evidence. Plausibility is 0 to 1. If you believe in X ', you can say that Bel (X') = 1, so the formula above is from Pls (X) = 0.
According to Giarratano and Riley, the Belief function can be formulated and shown in equation (1): And Plausibility is notated in equation (2) The Dempster-Shafer theory states that there is a frame of discourse denoted by the symbol (Θ). A frame of Decrement is the universe of discussion of a set of hypotheses so that it is often referred to as the environment shown in equation (3):

Results and Discussion
From the knowledge base collected, it can be grouped into 3 tables that are needed in analyzing. The tables used are the characteristic  Figure 1. There are three student learning styles and their explanations:

Visual Learning Style
Visual Learning Style emphasises visual acuity. That is, concrete evidence must be shown first so that they understand Learning styles like this rely on vision or see the evidence first and then can trust it. Some characteristics are typical of people who like this visual learning style. First is the need to see something information/lessons) visually to know it or understand it, second has a strong sensitivity to colour, third has an adequate understanding of artistic problems, fourth has difficulty in dialogue directly, fifth is too reactive to sound, six difficult to follow suggestions oral, the seventh often misinterpret words or sayings.

Auditory Learning Style
Auditory learning styles (Auditory Learners) rely on hearing to be able to understand and remember them. The characteristics of learning models such as these place hearing as the primary means of absorbing information or knowledge. That is, we must listen to, then we can remember and understand the information. The first character of people who have this learning style is that all information can only be absorbed through hearing, the second has difficulty absorbing information in written form directly, the third has difficulty writing or reading.

Kinesthetic Learning Style
Kinesthetic Learners (Kinesthetic Learners) require that the individual concerned touch something that provides certain information so that he can remember it. Of course, there are some characteristics of learning models like this that not everyone can do. The first character is placing the hand as the main recipient of information so that it can continue to remember it. Only by holding it, someone who has this style can absorb information without having to read the explanation. The weight of confidence in this system is the weight of the CF value inputted by the expert. In the system there are 20 questions, where for each answer in the question there are elements of learning style characters. Students can only choose one answer that matches their character. When students choose the answer, automatically the weight of the CF character value will be stored by the system. For more details, a manual calculation is provided in the case examples based on the answers from the questionnaire (attached) given to students presented in Table 4. below: From the results of all the manual calculations of the Dempster Shafer method with three input characteristics, the highest value obtained from M19 is that the student is in the category that matches the visual learning model with a confidence value of 95%. This value is in accordance with calculations through an expert system that has been developed in figure (4). It should be underlined that the confidence value chosen is based on the largest value of each density. This means that a low confidence score does not mean low confidence in the results of the diagnosis and vice versa.

User Interface Application Identification of Student Learning Styles
Student Learning Style Identification Application consists of two parts of the user (user interface) and administrator, in the user section, can identify student learning styles by entering the characters of students can be seen in Figure 3. while the administrator part contains input Criteria Data, Learning Style Data and Rule Data can be seen in Figure 4.

Conclusion
Based on the discussion in the previous chapter it can be concluded that the application of learning model identification using Demster Shafer Theory has been successfully built, the application can identify student learning styles by tracing the characters of students. Suggestions for future research need to be reviewed again related to the learning model and character of students, to see the comparison of identification accuracy levels can be compared using other algorithms, and so that it can be accessed easily by users can consider by developing an Android application.