ANALISIS PERBANDINGAN KLASIFIKASI DOKUMEN SKRIPSI MENGGUNAKAN SUPPORT VECTOR MACHINE, NAÏVE BAYES CLASIFIER, DAN K-NEAREST NEIGHBOR

SD Abdurrahman, NIM.: 18106050029 (2022) ANALISIS PERBANDINGAN KLASIFIKASI DOKUMEN SKRIPSI MENGGUNAKAN SUPPORT VECTOR MACHINE, NAÏVE BAYES CLASIFIER, DAN K-NEAREST NEIGHBOR. Skripsi thesis, UIN SUNAN KALIJAGA YOGYAKARTA.

Preview	Text (ANALISIS PERBANDINGAN KLASIFIKASI DOKUMEN SKRIPSI MENGGUNAKAN SUPPORT VECTOR MACHINE, NAÏVE BAYES CLASIFIER, DAN K-NEAREST NEIGHBOR) 18106050029_BAB-I_IV-atau-V_DAFTAR-PUSTAKA.pdf - Published Version Download (3MB) | Preview
	Text (ANALISIS PERBANDINGAN KLASIFIKASI DOKUMEN SKRIPSI MENGGUNAKAN SUPPORT VECTOR MACHINE, NAÏVE BAYES CLASIFIER, DAN K-NEAREST NEIGHBOR) 18106050029_BAB-II_sampai_SEBELUM-BAB-TERAKHIR.pdf - Published Version Restricted to Registered users only Download (4MB) | Request a copy

Abstract

UIN Sunan Kalijaga has a gallery of reading materials in the form of theses contained in the library repository. Of course, with this large number, the process of finding the right thesis will take a long time because you have to check it one by one. This is considered less effective and efficient so there must be a system that can facilitate this procession. By classifying the many thesis, it can be further divided into small parts which of course can make the search process more effective and efficient. The large number of theses will be divided into four themes, namely: Information Systems (SI), Computer Network Systems (SJK), Intelligent Systems (SC), and Software Engineering (RPL). However, in the realization itself, errors are still often found, by utilizing the error classification system, it can be reduced which will have a good impact in the future. From the results of research conducted using data from the last five years (from January 2017 until the research was made) there are 247 data, the data is divided into four themes, named: Information systems (SI), Computer Network Systems (SJK), Intelligent Systems (SC), and Software Engineering (RPL), with a total of 13, 25, 130, and 79 data, respectively. From the modeling using the data, Support Vector Machine gets an accuracy of 82%, then Naïve Bayes is 72%, and finally K-Nearest Neighbors is 84%. In addition, it can also be concluded that SVM is more sensitive to data outliers than the other two algorithms, due to the precision value in SVM in determining the least data theme, namely Network and Computer System

Share this knowledge with your friends :

Actions (login required)

View Item