Period Study Accuracy Prediction using Sequential Minimal Optimization Algorithm

Abstract

The study period is quite influential in the assessment of a university. The imbalance in the ratio of students to lecturers causes the quality of teaching and learning to decline, this is because one lecturer has to manage many students. Acquisition of accreditation scores and society's assumptions about higher education are also strongly influenced by the number of student graduations on time. Therefore, the prediction of the accuracy of the study period is needed as consideration for related parties to solve the problem of student learning delay. Sources of data in this study were taken from a database stored at the University of Surakarta, namely the Temporary Achievement Index with data of 209 instances and 5 attributes. The proposed method in this study is the Sequential Minimal Optimization algorithm. The validation method uses k-fold Cross-Validation with a value of K = 10. This method is compared with other methods such as naive Bayes, KNN, and Decision Tree. The results of this study, the proposed method can predict the accuracy of the study period quite well with the acquisition of accuracy of 88.52%. However, several other methods such as NaiveBayes obtained better accuracy of 90.91%, KNN of 91.86%, and Decision Tree of 96.65%. From the results of the comparison of these methods, the Decision Tree obtained the highest accuracy value. In future studies, researchers aim to enrich features in the prediction process. These features are related to student activities, such as student backgrounds, social activities, additional activities on campus and off-campus, and other aspects.