KLASIFIKASI KELULUSAN MAHASISWA MENGGUNAKAN METODE NAIVE BAYES DENGAN KOMPARASI PRAPEMROSESAN RANDOM OVERSAMPLING SERTA SELEKSI FITUR INFORMATION GAIN DAN FORWARD SELECTION (STUDI KASUS: FAKULTAS SAINS DAN TEKNOLOGI UIN SUSKA RIAU)

Dony Fahrudy, NIM.: 20206052006 (2022) KLASIFIKASI KELULUSAN MAHASISWA MENGGUNAKAN METODE NAIVE BAYES DENGAN KOMPARASI PRAPEMROSESAN RANDOM OVERSAMPLING SERTA SELEKSI FITUR INFORMATION GAIN DAN FORWARD SELECTION (STUDI KASUS: FAKULTAS SAINS DAN TEKNOLOGI UIN SUSKA RIAU). Masters thesis, UIN SUNAN KALIJAGA YOGYAKARTA.

Preview	Text (KLASIFIKASI KELULUSAN MAHASISWA MENGGUNAKAN METODE NAÏVE BAYES DENGAN KOMPARASI PRAPEMROSESAN RANDOM OVERSAMPLING SERTA SELEKSI FITUR INFORMATION GAIN DAN FORWARD SELECTION (STUDI KASUS: FAKULTAS SAINS DAN TEKNOLOGI UIN SUSKA RIAU)) 20206052006_BAB-I_IV-atau-V_DAFTAR-PUSTAKA.pdf - Published Version Download (3MB) | Preview
	Text (KLASIFIKASI KELULUSAN MAHASISWA MENGGUNAKAN METODE NAÏVE BAYES DENGAN KOMPARASI PRAPEMROSESAN RANDOM OVERSAMPLING SERTA SELEKSI FITUR INFORMATION GAIN DAN FORWARD SELECTION (STUDI KASUS: FAKULTAS SAINS DAN TEKNOLOGI UIN SUSKA RIAU)) 20206052006_BAB-II_sampai_SEBELUM-BAB-TERAKHIR.pdf - Published Version Restricted to Registered users only Download (8MB) | Request a copy

Abstract

The imbalance in the amount of data in each class and having high attribute dimensions in datasets, is often a problem in the classification process that can affect the algorithm's performance in the computational process, because there is an unbalanced amount of data in each class and irrelevant attributes must be processed, so it is necessary to do this. class imbalance techniques to balance the amount of data in each class and feature selection to reduce data complexity and irrelevant features. Therefore, this study uses the Random Oversampling (ROs) technique to overcome class imbalances, as well as two feature selection methods with Information Gain and Forward Selection algorithms which are compared to determine which feature selection method is superior, more effective, and more suitable. used. The results of feature selection are used to classify student graduation by building a classification model using the Naïve Bayes algorithm. The results showed an increase in the average accuracy of the Naïve Bayes method with no ROs preprocessing and feature selection, the use of ROs, the use of Information Gain with 3 selected features and Forward Selection with 2 selected features sequentially was 81.83%; 83.84%; 86.03% and 86.42%, so that there is an increase in accuracy of 4.2% from no pre-processing to Information Gain and 4.59% from no pre-processing to Forward Selection. Therefore, the best feature selection method is Forward Selection with 2 selected features (Ip Semester 8 and GPA), the use of ROs and both feature selections are proven to improve the performance of the Naïve Bayes method.

Share this knowledge with your friends :

Actions (login required)

View Item