S. Tamaela, Y. A. Lesnussa, V. Y. I. Ilwaru, A. M. Balami, Analysis of Support Vector Machine (SVM) Method and Simple Additive Weighting (SAW) Method in Making Decisions to Choose Specialization How to cite: S. Tamaela, Y. A. Lesnussa, V. Y. I. Ilwaru, and A. M. Balami, “Analysis of Support Vector Machine (SVM) Method and Simple Additive Weighting (SAW) Method in Making Decisions to Choose Specialization”, J. Mat. Mantik, vol. 6, no. 2, pp.104-112, October 2020. Analysis of Support Vector Machine (SVM) Method and Simple Additive Weighting (SAW) Method in Making Decisions to Choose Specialization Stevanny Tamaela1, Yopi Andry Lesnussa2, V. Y. I. Ilwaru3, A. M. Balami4 1Universitas Pattimura Ambon, annytamaela@gmail.com 2Universitas Pattimura Ambon, yopi_a_lesnussa@yahoo.com 3Universitas Pattimura Ambon, vennilwaru007@gmail.com 4Universitas Pattimura Ambon, abdulmalikbalami@gmail.com doi: https://doi.org/10.15642/mantik.2020.6.2.104-113 Abstrak. Peminatan peserta didik adalah suatu pembelajaran berbasis minat peserta didik sesuai kesempatan belajar yang ada dalam satuan pendidikan. Penyelenggaraan pendidikan dalam satuan pendidikan di SMA berdasarkan kurikulum 2013 terdapat program penentuan peminatan bagi peserta didik SMA yang dilaksanakan di kelas X. Peminatan dalam kurikulum 2013 di SMA adalah kelompok peminatan IPA dan IPS. Penelitian ini menggunakan metode Support Vector Machine (SVM) dan metode Simple Additive Weighting (SAW) yang bertujuan untuk membandingkan tingkat keakuratan tiap metode dalam Sistem Pengambilan Keputusan (SPK) peminatan IPA dan IPS pada SMA Negeri 1 Ambon. Dari hasil penelitian diperoleh hasil pemilihan peminatan dari metode SAW berbeda dengan data riil, sedangkan hasil dari metode SVM menunjukan hasil yang sama dengan pemilihan peminatan riil di SMA Negeri 1 Ambon. Kata kunci: Kurikulum 2013; Support vector machine; Simple additive weighting Abstract. The specialization of students is a learning based on the interests of students according to learning opportunities that exist in educational units. Providing education in high school education units based on the 2013 curriculum there is a program for determining specialization for high school students held in class X. Specialization in the 2013 curriculum in high schools is the specialization group for Natural Sciences and Social Sciences. This study uses the Support Vector Machine (SVM) method and the Simple Additive Weighting (SAW) method which aims to compare the accuracy of each method in Decision Making (SPK) specialization program in the Natural Science and Social Sciences at SMA Negeri 1 Ambon. From the research results, the results of the specialization selection from the SAW method differ from the real data, while the results of the SVM method show the same results as the selection of real specialization in SMA Negeri 1 Ambon Keywords: 2013 Curriculum; Support vector machine; Simple additive weighting Jurnal Matematika MANTIK Vol. 6, No. 2, October 2020, pp. 104-113 ISSN: 2527-3159 (print) 2527-3167 (online) mailto:vennilwaru007@gmail.com3 http://u.lipi.go.id/1458103791 S. Tamaela, Y. A. Lesnussa, V. Y. I. Ilwaru, A. M. Balami, Analysis of Support Vector Machine (SVM) Method and Simple Additive Weighting (SAW) Method in Making Decisions to Choose Specialization 105 1. Introduction The curriculum is the most important part of the National education system, whose existence has existed since the beginning of the existence of National education [1]. The curriculum is a plan about the formation of abilities and character of children based on a standard [2], 2013 curriculum is a curriculum that simplifies[3] and thematic-integrative, adding rainy hours to encourage students . One of the forms is class division specialization at the Senior High School level [4]. Specialization aims to provide opportunities for students to develop their interests in a group of subjects by their scientific interests in higher education[5] and develop their interest in a particular discipline or skill[6]. According to [6] Decision-Making System is the process of choosing between two or more alternative actions to achieve goals or objectives. Several studies have concluded that various decision- making methods can determine reliable results. However, in this case, two methods will be applied including (1) the Support Vector Machine (SVM) Algorithm, the SVM method is a technique for making predictions, both in the case of classification and regression [7]. SVM was introduced by Boser, Guyon, Vanpik, and was first presented in 1992 [7] at the Annual Workshop on Computational Learning Theory. The basic concept of SVM is a harmony of computational theories that existed decades before, such as the hyperplane margin introduced by Aronszajn in 1950 However, up until 1992 there had never been any attempt to assemble these components. The basic principle of SVM is a linear classifier [8] and subsequently developed to work on non-linear problems by incorporating the concept of kernel tricks in high-dimensional space [9]. The main purpose of SVM is to increase speed in training and testing so that SVM can be used for large data (2) Simple Additive Weighing (SAW) method, the basic concept of the SAW method is to find the weighted sum of the performance ratings for each alternative on all attributes [10][11]. SAW can carry out a more precise assessment because it is based on criteria values [12] and weights that have been determined and can select the best alternative. Larger values indicate that alternatives are preferred [13]. The Simple Additive Weighting (SAW) method is recommended to solve the selection problem in a multi-process decision-making system [14]. Simple Additive Weight (SAW) method is a method that is widely used in decision making that has many attributes [15]. In this paper, we will compare the accuracy of SVM methods and SAW methods in decision making system to determine the specialization in SMA Negeri 1 Ambon. 2. Methods The type of research used is a case study, by comparing the SVM method with the SAW method to determine the choice of interest in tenth grade students. The material used in this study is secondary data obtained from SMA Negeri 1 Ambon. Secondary data taken from SMA Negeri 1 Ambon in the form of a value criterion as a measure of interest selection include the initial test scores in SMA 1 or a comparative value to determine a Major Course in tenth grade students. The Simple Additive Weighting (SAW) work process flowchart can be seen in the following figure: Jurnal Matematika MANTIK Volume 6, No. 2, October 2020, pp.104-113 106 Figure 1. Simple Additive Weighting (SAW) work process flow chart In detail the workflow diagram of the Support Vector Machine (SVM) can be seen in the following figure: Figure 2. The work process of Support Vector Machine (SVM) SVM linear search for hyperplane with the largest margin, known as the Maximum Marginal Hyperplane (MMH). Based on the Lagrangian formulation mentioned, MMH can be rewritten as a boundary decision: 𝑑(𝑋𝑇) = ∑ 𝑦𝑖𝛼𝑖𝑋𝑖𝑋 𝑇 + 𝑏0 𝑙 𝑖=1 (1) START Enter data from each criterion on each alternative Criteria Weight Make a Decision Matrix Normalization of the decision matrix Multiply the Decision Matrix by Criteria Weight Preference for each alternative FINISH START Divide Data into Training and Testing Yes SVM Classification if Score > 0 No Natural Science (IPA) interest Social Science (IPS) interest Classification Results FINISH S. Tamaela, Y. A. Lesnussa, V. Y. I. Ilwaru, A. M. Balami, Analysis of Support Vector Machine (SVM) Method and Simple Additive Weighting (SAW) Method in Making Decisions to Choose Specialization 107 where 𝑦𝑖 is the label vector support class 𝑋𝑖. 𝑋 𝑇 is a test tuple; 𝛼𝑖 and 𝑏0 are numerical parameters that are automatically determined by the SVM algorithm, and l is the number of Support Vector. 3. Results and Discussion 3.1. SAW Method For this research method, the variables used for data are the National Examination Score (C1), Entrance Test Score (C2), Psychological Value (C3). The amount of data in this study amounted to 50 data. The interest or major divide in 2 part, such as: Natural sciences department (IPA) and Social sciences department (IPS). Table 1. Research Data No Ne score (rek) Initial Score Value Psychology Score Final Score Major/ interest 1 82 35 106.7 74.57 IPA 2 80.2 34 86.2 66.8 IPS 3 80 34 100 71.33 IPA 4 81.5 34 94.4 69.97 IPS 5 80 34 93.9 69.30 IPS 6 85.5 33 111.8 76.77 IPA 7 84.5 33 111.9 76.47 IPA 8 80 33 86.1 66.37 IPS 9 81.5 33 109.6 74.70 IPA 10 80.8 32 106.3 73.03 IPA 11 80 32 97 69.67 IPS 12 83 32 106.7 73.90 IPA 13 82.5 32 92.6 69.03 IPS 14 81.9 31 107.7 73.53 IPA 15 80.5 31 92.6 68.03 IPS 16 80 31 98.7 69.90 IPS 17 81.5 31 83.3 65.27 IPS 18 80 31 91.3 67.43 IPS 19 85.5 30 90.6 68.70 IPS 20 84.5 30 101.3 71.93 IPA 21 80 30 89.7 66.57 IPS 22 80 30 93.6 67.87 IPS 23 83 30 88.3 67.10 IPS 24 82.5 30 111.4 74.63 IPA 25 81.9 30 97.7 69.87 IPS 26 80.5 30 109.6 73.37 IPA 27 80 30 104.5 71.50 IPA 28 81.5 30 100 70.50 IPA 29 83 30 91.1 68.03 IPS 30 82.5 30 103.9 72.13 IPA 31 81.9 30 100 70.63 IPA 32 80.5 28 104.2 70.90 IPA 33 83.9 28 106 72.63 IPA Jurnal Matematika MANTIK Volume 6, No. 2, October 2020, pp.104-113 108 No Ne score (rek) Initial Score Value Psychology Score Final Score Major/ interest 34 84 28 92.9 68.30 IPS 35 80.5 28 87.7 65.40 IPS 36 80.4 28 86.9 65.10 IPS 37 81.3 28 111.9 73.73 IPA 38 81.9 28 92.5 67.47 IPS 39 81 28 86.8 65.27 IPS 40 80.2 28 105 71.07 IPA 41 82 27 109.9 72.97 IPA 42 83.6 27 110.5 73.70 IPA 43 80 27 93.8 66.93 IPS 44 82.5 27 94.9 68.13 IPS 45 81.6 27 91.3 66.63 IPS 46 85 27 106.6 72.87 IPA 47 82.9 27 106 71.97 IPA 48 83.5 27 94.7 68.40 IPS 49 80 27 93.3 66.77 IPS 50 80 27 111.1 72.70 IPA In this research process, the calculation process will be carried out using sample data from students who will register at SMAN 1 Ambon, using the SAW method. Alternative data used in the study are in table 2. Table 2. Alternative Data Alternatives Ai A1 Natural sciences (IPA) A2 Social Sciences (IPS) The criteria used to determine each alternative above are the National Examination Score, Initial Test Score, and Psychological Test Score. Departmental target data is divided into 3 groups with the following conditions: Department of Natural Sciences for a value of ≤ 85 with very clever provisions Department of Neutral for a Value of 70 ≤ y ≤ 80 with the provisions of the Average Ministry of Social Sciences for Low value of 70 with a low provision Note: y = Major Based on subject criteria, it is determined that preference (W) is = {30, 30, 40}. After preference weights are determined, a matrix is made based on the previous weighting table. Then, the X matrix is normalized, based on the equation of the SAW method to obtain an R normalized matrix, then the normalized matrix R is multiplied by W which is the weight of the predetermined preference. S. Tamaela, Y. A. Lesnussa, V. Y. I. Ilwaru, A. M. Balami, Analysis of Support Vector Machine (SVM) Method and Simple Additive Weighting (SAW) Method in Making Decisions to Choose Specialization 109 Table 3. C ++ Software Results Table No Rij UN Rij TA Rij SP Vij Recommendation 1 0.75 1.00 0.67 79.17 IPA 2 0.75 1.00 0.33 65.83 IPS 3 0.75 1.00 0.33 65.83 IPA* 4 0.75 1.00 0.33 65.83 IPS 5 0.75 1.00 0.33 65.83 IPS 6 1.00 1.00 1.00 100.00 IPA 7 1.00 1.00 1.00 100.00 IPA 8 0.75 1.00 0.33 65.83 IPS 9 0.75 1.00 0.67 79.17 IPA 10 0.75 1.00 0.67 79.17 IPA 11 0.75 1.00 0.33 65.83 IPS 12 0.75 1.00 0.67 79.17 IPA 13 0.75 1.00 0.33 65.83 IPS 14 0.75 1.00 0.67 79.17 IPA 15 0.75 1.00 0.33 65.83 IPS 16 0.75 1.00 0.33 65.83 IPS 17 0.75 1.00 0.33 65.83 IPS 18 0.75 1.00 0.33 65.83 IPS 19 1.00 0.50 0.33 58.33 IPS 20 1.00 0.50 0.33 58.33 IPA* 21 0,75 0.50 0.33 58.33 IPS 22 0.75 0.50 0.33 58.33 IPS 23 0.75 0.50 0.33 58.33 IPS 24 0.75 0.50 1.00 77.50 IPA 25 0.75 0.50 0.33 50.83 IPS 26 0.75 0.50 0.67 64.17 IPA 27 0.75 0.50 0.33 50.83 IPA 28 0.75 0.50 0.33 50.83 IPA* 29 0.75 0.50 0.33 50.83 IPS 30 0.75 0.50 0.33 50.83 IPA* 31 0.75 0.50 0.33 50.83 IPA* 32 0.75 0.50 0.33 50.83 IPA* 33 0.75 0.50 0.67 64.17 IPA* 34 0.75 0.50 0.33 50.83 IPS 35 0.75 0.50 0.33 50.83 IPS 36 0.75 0.50 0.33 50.83 IPS 37 0.75 0.50 1.00 77.50 IPA 38 0.75 0.50 0.33 50.83 IPS 39 0.75 0.50 0.33 50.83 IPS 40 0.75 0.50 0.33 50.83 IPA 41 0.75 0.50 0.67 64.17 IPA* 42 0.75 0.50 1.00 77.50 IPA Jurnal Matematika MANTIK Volume 6, No. 2, October 2020, pp.104-113 110 No Rij UN Rij TA Rij SP Vij Recommendation 43 0.75 0.50 0.33 50.83 IPS 44 0.75 0.50 0.33 50.83 IPS 45 0.75 0.50 0.33 50.83 IPS 46 1.00 0.50 0.67 71.67 IPA 47 0.75 0.50 0.67 64.17 IPA* 48 0.75 0.50 0.33 50.83 IPS 49 0.75 0.50 0.33 50.83 IPS 50 0.75 0.50 1.00 77.50 IPA According to Table 3, 𝑅𝑖𝑗 UN, 𝑅𝑖𝑗 TA, 𝑅𝑖𝑗 SP, and 𝑉𝑖𝑗 representing respectively, the weight of National Examination, Entrance Test, Psychological Value and the final score. From the final score or final value will use to determine the decision making to choose the specialization in natural sciences department or social sciences department. 3.2. SVM Method For this research method, the variables used for data are the National Examination Score (X1), Entrance Test Score (X2), Psychological Value (X3). While the target or output is Class or Major Course (Y). The data used are the same as Table 1. This study uses the linear SVM method with MATLAB software to conduct data analysis. A comparison of training and testing data used is 70:30 from each grouping that has been determined. The results of the software for grouping the training data are as follows: Table 4. Software Results for Testing Data No Original Data svmStruct1 svmStruct2 svmStruct3 Software Results 1 0 0 0 0 0 2 0 1 0 0 0 3 0 1 0 0 0 4 0 0 0 0 0 5 0 0 0 0 0 6 0 0 0 0 0 7 0 0 0 0 0 8 0 1 0 0 0 9 1 1 1 1 1 10 1 1 1 1 1 11 1 1 1 1 1 12 1 1 1 1 1 13 1 0 1 1 1 14 1 1 1 1 1 15 1 0 1 1 1 16 1 1 1 1 1 From Table 4, it can be seen the SVM method classification shows the output with 100% accuracy for training data. And also, class 0 is a group of Natural Sciences Department) and class 1 is a group of Social Sciences Department. This can be seen from the comparison between the target data and the output target. svmStruct1, svmStruct1, S. Tamaela, Y. A. Lesnussa, V. Y. I. Ilwaru, A. M. Balami, Analysis of Support Vector Machine (SVM) Method and Simple Additive Weighting (SAW) Method in Making Decisions to Choose Specialization 111 svmStruct1 each representing national examination, entrance test, and psychological value. With Hyperplan obtained from each class as follows: Figure 3. Hyperplane for svmStruct1 Based on Figure 3, it can be explained that the svmStruct1 data shows that the data in the two classes are not completely separate can be seen from several red circles whose distribution is around the green circle area. Figure 4. Hyperplane for svmStruct2 Based on Figure 4, it can be explained that the data in svmStruct1 is correctly classified in class 0 (a group of Natural Sciences Department) or class 1 (a group of Social Sciences Department) while the Support Vector is a circle within the circle. Figure 5. Hyperplane for svmStruct3 Jurnal Matematika MANTIK Volume 6, No. 2, October 2020, pp.104-113 112 Based on Figure 5, it can be explained that all svmStruct3 data is properly classified into class 0 (which is the Natural Science Department) or class 1 (which is the Social Sciences Department), while the Support Vector is the circle within the circle. By the purpose of this study to see the level of accuracy between the SAW method and the SVM method, so the results obtained namely the SAW method are found. From the results in Table 3 above it can be seen that 12 students can enter the natural science (IPA) class and 38 in the social science (IPS) class. This shows that there are differences in decision criteria in the Determination a Major Course according to the results of calculations with the SAW method in Table 3, with the results of student selection in the data in Table 1. Some of the different data can be seen in data 3, 20, 28, 30, 31, 32, 33, 41, 47 which are different from the actual data in Table 1. From these different data, the results of the SAW calculation are more suitable for the IPS interest class, but in the real data, it is in the IPA interest class. As for the SVM method, the results of the software output matched with the real data are seen that 100% of the output data is the same as the real data. This shows for the SVM method of Determining a Major Course carried out by the school according to the ability of children. 4. Conclusions From the results and discussion can be conclude that, the results of research by the SAW method there are some students who have low criteria but are in the interest or department of Natural Sciences and vice versa. However, based on the results of research with the SVM method there are classifications of interest or majors that are different from the interests or classifications obtained are imperfect classifications. References [1] M. Ali, “Implementasi Kurikulum Pendidikan Nasional 2013,” J. Pedagog., Vol. 2(2), pp. 49-60, 2013. [2] M. Lestari, “Implementation of Citizenship Character Formation by the Study of Civic Education on Senior High School in The District of Bantul,” E-CIVICS, Vol. 5 (6), 2016. [3] Suyatmini, “Implementasi Kurikulum 2013 pada Pelaksanaan Pembelajaran Akuntansi di Sekolah Menengah Kejuruan,” J. Pendidik. Ilmu Sos., Vol. 20(1), pp.60- 68, 2017. [4] Subandi, “Pengembangan Kurikulum 2013,” J. Pendidik. dan Pembelajaran Dasar, Vol. 1(1), pp. 18-36, 2014. [5] B. L. Julien, L. Lexis, J. Schuijers, T. Samiric, and S. McDonald, “Using capstones to develop research skills and graduate capabilities: A case study from physiology,” J. Univ. Teach. Learn. Pract., Vol. 9(3), 2012. [6] Direktorat Pembinaan SMA, Modul Pelatihan Implementasi Kurikulum 2013 SMA Tahun 2018. pp. 1-62, 2018. [7] Y. Lin, H. Tseng, and C. Fuh, “Using Support Vector Machine,” Image Process., pp.123-130, 2003. [8] M. A. Oskoei and H. Hu, “Support vector machine-based classification scheme for myoelectric control applied to upper limb,” IEEE Trans. Biomed. Eng., Vol. 55(8), pp. 1956-1965, 2008, doi: 10.1109/TBME.2008.919734. [9] S. Vijayakumar and S. Wu, “Sequential Support Vector Classifiers and Regression,” Proceedings of International Conference on Soft Computing (SOCO ‘99), Vol. S. Tamaela, Y. A. Lesnussa, V. Y. I. Ilwaru, A. M. Balami, Analysis of Support Vector Machine (SVM) Method and Simple Additive Weighting (SAW) Method in Making Decisions to Choose Specialization 113 1999(619), pp. 610-619, 1999. [10] M. Elistri, J. Wahyudi, and R. Supardi, “Penerapan Metode SAW Dalam Sistem Pendukung Keputusan Pemilihan Jurusan Pada Sekolah Menengah Atas Negeri 8 Seluma,” J. Media Infotama Penerapan Metod. SAW…ISSN, Vol. 10(2), pp. 1858- 2680, 2014. [11] S. H. Sahir, R. Rosmawati, and K. Minan, “Simple Additive Weighting Method to Determining Employee Salary Increase Rate,” Ijsrst, Vol. 3(8), pp. 42-48, 2017. [12] E. Roszkowska and D. Kacprzak, “The fuzzy SAW and fuzzy TOPSIS procedures based on ordered fuzzy numbers,” Inf. Sci. (Ny)., Vol. 369, pp. 564-584, 2016, doi: 10.1016/j.ins.2016.07.044. [13] S. E. Widodo Sri, S. Lutfi, and Solikhin, “Sistem Pendukung Keputusan Penilaian Kinerja Karyawan Menggunakan Metode Simple Additive Weighting (SAW) Pada Pt. Indonesia Steel Tube Work,” Sist. Inf., 2014. [14] F. Fitriani, “Sistem Pendukung Keputusan Penentuan Jenis Rambut Manusia Dengan Menerapkan Metode Simple Additive Weighting (SAW),” Pelita Inform. Budi Darma, 2015. [15] Ratnasari and T. Susilowati, “Sistem Pendukung Keputusan Kelayakan Pengajuan Kredit Sepeda Motor pada Dealer Tunas Dwipa Matra Gadingrejo Menggunakan Metode SAW,” STMIK Pringsewu Lampung, pp. 442-448, 2016.