Author : Mustakim, Reysa Hastarimasuci, Petir Papilo, Zarkasih, Zaitun, Alwis Nazir
Publish : The 1st Workshop on Environmental Science, Society, and Technology. Journal of Physics: Conference Series 1363 (2019)
Appropriate student’s major placement in high school can help students to better improve their academic achievement. There are many variables which must be considered to determine the student’s majors, such as: Gender, Interests, Intelligence Quotient (IQ); Four subjects in Junior High School (JHS), average junior high school grades, matriculation score of four subjects, and average rate of matriculation. The number of variables used in the selection, causes some weaknesses among them, such as the complexity of variable, the inefficiency of variable and the existence of some variables which is only as an addition without having a significant contribution. This study aims to reduce the number of these variables so it will become easier to analyze and to be applied. The Process of reduction was done by combining experiments with Predefined attributes. A total of ten combinations were attempted using KNearest Neighbor (K-NN) and Naïve Bayes Classifier (NBC) which then was measured by Confusion Matrix accuracy. The experimental result showed that the combination of variables which produce the best accuracy were the 9th and 10th experiment with variable matriculation, interest, and IQ, and an accuracy of 96.77% from K-NN also 98.38% from NBC. By combining both algorithms, 99.87% of maximum accuracy was obtained from those three variables. New information which can be extracted from this research is that there are only three important variables to determine major placement in Senior High School, Average Scores of Matriculation, Interests and IQ followed by four supporting variables such as the scores of Mathematic, Physics, English and Economics in Matriculation.
The calculation of NBC algorithm applied for student major classification in Islamic of Senior High School 2 Pekanbaru Model is able to produce the highest accuracy of 98.38% at S-9, it is higher than K-NN 96.77% at S-10. S-9 and S-10 contain combination of main variable, the average of matriculation scores with interest and interest with IQ. So it can be concluded that to determine the major placement, the most important variables which need to be the main references are the average scores of Matriculation, Interests and IQ. The three variables were also combined to two algorithms, K-NN and NBC. Both algorithms were compared by applying the accuracy greater than 90% which obtained the accuracy of K-NN to NBC by 99.87% and NBC to K-NN equal to 99.03%. The result of this research in general can be stated that there are only 3 main variables and 4 matriculation scores as supporting variables that can be used as reference in the major placement, not thirteen attributes as previously proposed.