Prediction Of Student Graduation Using The K-Nearest Neighbor Method Case Study in Politeknik Negeri Tanah Laut

Abstract

Tanah Laut State Polytechnic as one of the universities in Indonesia has definitely paid attention to the quality of its students. One way is to predict student graduation. Graduation predictions can help study programs and academic supervisors review and pay special attention to students, especially students who are predicted to not graduate on time. Realizing one way to pay attention to the quality of students can be realized by creating a Student Graduation Prediction system using the Web-Based K-Nearest Neighbor (KNN) Method. The K-Nearest Neighbors method is an object classification method based on training data by finding the nearest neighbor value to determine the class of the new data. In the Student Graduation Prediction using the K-Nearest Neighbor Method, there is a section that can process training data, test data, the process of calculating student graduation predictions, and displaying the results obtained from the KNN calculation which has two classification classes, namely graduated and not passed. Based on the results of the study, it was found that KNN with different k values obtained different levels of accuracy, data testing with a value of k=1 obtained an accuracy rate of 83.33%, the value of k=2 obtained an accuracy rate of 79.17%, the value of k=3 to k= 8 obtained an accuracy rate of 95.83%, and the values of k=9 and k=10 obtained an accuracy rate of 91.67%. It can be concluded that the test with a value of k=3 to k=8 obtained the best or highest level of accuracy.