Accreditation classification of junior high school in Indonesia using K-nearest neighbor, logistic regression, classification tree

In order to improve the quality of education in Indonesia, the government created a program called accreditation which aims to improve the quality of education based on the National Education Standards (SNP) which pays attention to 8 educational standards. In addition, accreditation aims to motivate schools/madrasah to continue to gradually improve the quality of education at the district/city, provincial, national, and even regional and international levels as well as to help identify schools/madrasah in the context of providing assistance. A school/madrasah that meets the eight characteristics according to the SNP is accredited A, for those who lack/have not met the accreditation other than A. Executive Director of the National Accreditation Board for Higher Education (BAN-PT). This study specifically compares three supervised machine learning methods, namely k-nearest neighbor (kNN), logistic regression and classification tree. The results of the accuracy of the three methods show that there is no significant difference in the accuracy, the accuracy produced is almost the same for the three methods, 79-80%. Based on the measurement scale on the independent variable data, namely numeric and categorical, it will be inconvenient to classify using the kNN method because we have to standardize the data first. Meanwhile, logistic regression and classification tree, the difference in measurement scale in the data is not a problem. However, the author prefers to use a classification tree because by using a classification tree we can see the classification process and also know what variables are used as factors that distinguish the response variables.