PENERAPAN MACHINE LEARNING UNTUK MENGATASI KETIMPANGAN DATA DALAM MENENTUKAN KLASIFIKASI UANG KULIAH TUNGGAL (UKT)

Abstract

Single Tuition Fee or UKT is a tuition fee borne by students every semester. Payment is made every time students enter a new semester while studying at tertiary institutions. One of the state universities in Indonesia that has implemented the UKT payment system is Politeknik Negeri Ujung Pandang. Based on observations, the purchase of UKT is still done manually, so it has the potential to produce decisions that are not on target. This study was classified based on the amount of single student tuition fees using the smote method and without smote. The algorithm used in classifying is SVM, Decision Tree, Random Forest. The data used is 985 UKT data in 2021. Based on experiments that have been carried out with the Random Forest algorithm, it has the best performance compared to the SVM and Decision Tree algorithms. The proportion of results obtained before being hit is accuracy of 84.75%, precision of 79.22%, recall of 81.15% and F1 score of 80.17%. Whereas after applying smote it increased with an accuracy proportion of 98.9%. So it can be interpreted that the best algorithm used in classification is the Random Forest algorithm by applying smote.