Comparative Analysis Of Performance Levels Of Svm And Naïve Bayes Algorithm For Lifestyle Classification On Twitter Social Media

Abstract

Lifestyle is how individuals express themselves through their activities and interests and utilize their financial resources and available time. Twitter is a social network platform that allows people to express opinions and directly criticize various topics, including the recently widely discussed lifestyle topics. Topic classification on Twitter is central in facilitating the search, recommendation, and management of relevant content for users. This research aims to analyze public sentiment regarding lifestyle using 11,000 pieces of data with the keywords "concert", "watching films", "smoking", and others related to lifestyle. Research data is labeled according to the sentiment of public opinion towards lifestyle. Negative polarity for data that has the context of "underestimating", "insulting", "sarcastic", and "feeling sad". Positive polarity for data that has the context of "grateful", "praying", "feeling happy", and "encouraging". Neutral polarity for data that has the contexts “ask”, “predict”, and “feel surprised”. Next, the data enters the pre-processing stage, which consists of case-folding, tokenization, stopword removal, and lemmatizing. The analysis continues by dividing the data into training and test data with a ratio of 70%:30%. Sentiment analysis uses an algorithm Support Vector Machine (SVM) and Naïve Bayes Classifier (NBC). The analysis results show that the SVM algorithm provides better classification than NBC. In this case, the SVM algorithm produces accuracy, precision, recall, and value F1-Score the same, namely 61%.