Perbandingan Akurasi Algoritma Naïve Bayes dan Support Vector Machine Untuk Sentimen Analisis Komentar Pengguna Twitter (Studi Analisis: Peran Pemerintah Pada Penanganan Covid-19)
Total View This Week0
Total View This Week0
Institusion
Institut Teknologi Telkom Purwokerto
Author
Tasya Shabrina, Salsabillah
Subject
T Technology (General)
Datestamp
2022-07-15 04:19:03
Abstract :
Sentiment analysis on this study was conducted to classify Twitter users' comments on the role of goverment on andling COVID-19. The data used in this study is tweet data containing positive and negative comments on the goverment's role in handling COVID-19 with several uses of keywords including covid, covid 19, pandeic, goverment and PSBB for seacrhing and collecting data. An algorithm is need to analyze the existing dtaa. The algorithms used in this research are Support Vector Machine (SVM) and Naive Bayes Algorthms. WHere SVM is an algorithm with the concept of finding the best Hyperplane while the Naive bayes Algorithm for classifying data groupings the refers to the concepts of Probability and Statistics. The final result of this research is to compare the accuracy values of the two algorithms. The process carried out in this study begins with data labeling to determine positive and negative comments then the pre-processing and feature extraction process with TF-IDF to clean data, remove data from noise, and facilitate weighting. The dataset is divided into 80% training dataset and 20% test dataset. Then enter the calculation to determine the accuracy of each algorithm using the Confusion Matrix. After the preprocessing and word weighting stages using TF-IDF are known, the accuracy value obtained by the SVM algorithm is 83% while the NAive Bayes algorithm is 80%. After mesuring the model performance on both algorithms. The precision of the Naive Bayes algorithm is 80% with 80% recall and the SVM algorithm precision is 87% with 78% recall. This proves that the SVM algorithm is able to produce a better accuray value than the Naive Bayes algorithm in clasifying sentiment analysis of Twitter user comments on the government's role in handling COVID-19. Keywords: Accuracy, COVID-19, Classfication, Nave Bayes, SVM
Sentiment analysis on this study was conducted to classify Twitter users' comments on the role of goverment on andling COVID-19. The data used in this study is tweet data containing positive and negative comments on the goverment's role in handling COVID-19 with several uses of keywords including covid, covid 19, pandeic, goverment and PSBB for seacrhing and collecting data. An algorithm is need to analyze the existing dtaa. The algorithms used in this research are Support Vector Machine (SVM) and Naive Bayes Algorthms. WHere SVM is an algorithm with the concept of finding the best Hyperplane while the Naive bayes Algorithm for classifying data groupings the refers to the concepts of Probability and Statistics. The final result of this research is to compare the accuracy values of the two algorithms. The process carried out in this study begins with data labeling to determine positive and negative comments then the pre-processing and feature extraction process with TF-IDF to clean data, remove data from noise, and facilitate weighting. The dataset is divided into 80% training dataset and 20% test dataset. Then enter the calculation to determine the accuracy of each algorithm using the Confusion Matrix. After the preprocessing and word weighting stages using TF-IDF are known, the accuracy value obtained by the SVM algorithm is 83% while the NAive Bayes algorithm is 80%. After mesuring the model performance on both algorithms. The precision of the Naive Bayes algorithm is 80% with 80% recall and the SVM algorithm precision is 87% with 78% recall. This proves that the SVM algorithm is able to produce a better accuray value than the Naive Bayes algorithm in clasifying sentiment analysis of Twitter user comments on the government's role in handling COVID-19. Keywords: Accuracy, COVID-19, Classfication, Nave Bayes, SVM
Download
book
BibTex
Latex, Jabref
cloud_download
Original Resource
url resource Institution
