Comparison of Z-Score, Min-Max and No Normalization Methods using Support Vector Machine Algorithm to Predict Student^s Timely Graduation
Muhammad Sholeh1, b), Erna Kumalasari Nurnawati2,a)

Author Affiliations
1,2 Department of Informatics, Faculty of Information Technology and Business
Institut Sains & Teknologi AKPRIND Yogyakarta, Indonesia.


Author Emails
a) Corresponding author: ernakumala[at]akprind.ac.id
b)muhash[at]akprind.ac.id

Abstract

One indicator of the success of the higher education system is the timely graduation of students. Students who take undergraduate programs are declared to graduate on time if students can study for less than or equal to eight semesters. The success of graduation time can be monitored from the beginning of the semester by looking at the passing of courses in each semester. Normalization is done so that the resulting model has maximum accuracy. This study aims to build the best Classification model using Support Vector Machine (SVM) algorithm. That can predict students^ timely graduation by comparing the data normalization process, namely with the Z-Score, Min-Max and without normalization methods. The datasheet is taken from data on student study results in each semester in ten study programs at Institut Sains & Teknologi AKPRIND class 2017 as many as 267 data with 19 attributes. The model was developed using achievement index data from first to sixth semester. The recommended model is selected from the maximum accuracy results. The results showed that the classification model with the SVM algorithm using Z-score normalization produced the highest accuracy with an accuracy value of 83%. That is, the recommended model is a model using Z-Score normalization.

Keywords: Support Vector Machine, Classification, normalization methods

Topic: Big Data and Analytics