STUDI PERBANDINGAN METODE ANALISIS NAIVE BAYES CLASSIFIER DENGAN SUPPORT VECTOR MACHINE UNTUK ANALISIS SENTIMEN (STUDI KASUS: TWEET BERBAHASA INDONESIA TENTANG COVID-19)

Ahmad Nur Fauzi, NIM.: 16650029 (2020) STUDI PERBANDINGAN METODE ANALISIS NAIVE BAYES CLASSIFIER DENGAN SUPPORT VECTOR MACHINE UNTUK ANALISIS SENTIMEN (STUDI KASUS: TWEET BERBAHASA INDONESIA TENTANG COVID-19). Skripsi thesis, UIN SUNAN KALIJAGA YOGYAKARTA.

Preview	Text (STUDI PERBANDINGAN METODE ANALISIS NAIVE BAYES CLASSIFIER DENGAN SUPPORT VECTOR MACHINE UNTUK ANALISIS SENTIMEN (STUDI KASUS: TWEET BERBAHASA INDONESIA TENTANG COVID-19)) 16650029_BAB-I_IV-atau-V_DAFTAR-PUSTAKA.pdf - Published Version Download (4MB) | Preview
	Text (STUDI PERBANDINGAN METODE ANALISIS NAIVE BAYES CLASSIFIER DENGAN SUPPORT VECTOR MACHINE UNTUK ANALISIS SENTIMEN (STUDI KASUS: TWEET BERBAHASA INDONESIA TENTANG COVID-19)) 16650029_BAB-II_sampai_SEBELUM-BAB-TERAKHIR.pdf - Published Version Restricted to Registered users only Download (6MB) | Request a copy

Abstract

Sentimen analysis is a technique for detecting opinions about a subject (for example individuals, organizations or products) in a data set. Community life today supports the emergence of social interaction through social media which is a suggestion in conveying positive and negatif opinions. In this study using 2 methods, namely Naïve Bayes Classifiers (NBC) and Support Vector Machine (SVM) to classify sentimens. A total of 9015 data were used in the analysis process with TF-IDF weighting techniques and Term Frequency. Evaluation uses 2 ways, that is confusion matrix and K-fold cross validation. With the confusion matrix, SVM has an accuracy value of 83%, a precision value of 83%, a recall value of 98.8%, and a specificity value of 9%. While NBC has an accuracy value of 82.3%, a precision value of 82.3%, a recall value of 99.7%, and a specificity value of 3.4% with TFIDF. Then, SVM has an accuracy value of 82.8%, a precision value of 83%, a recall value of 99%, and a specificity of 8%. While NBC has an accuracy value of 80.1%, a precision value of 85.7%, a recall value of 90%, and a specificity of 29% with TF. K-fold cross Validation with TF-IDF weighting, SVM accuracy value of 82.6%, precision value of 82.9%, recall rate of 98.8%, and specificity of 10.9%. NBC accuracy is 82%, precision value is 82%, recall rate is 99.6%, and specificity is 4.8%. SVM Term Frequency accuracy value is 82.5%, precision value is 82.6%, 99.1% recall rate, and specificity is 7%. NBC accuracy value is 79.8%, precision value is 85.9%, recall rate is 89.2%, and specificity is 32%. NBC data processing time is faster than SVM with NBC under 5 seconds while SVM is above 500 seconds.

Share this knowledge with your friends :

Actions (login required)

View Item