Analisis Sentimen Terhadap Rangka E-SAF Honda Pada Media Sosial X Dengan Algoritma Naïve Bayes

Abstract

Motorcycles are the best vehicles for traveling when traffic is heavy because motorcycles allow people to save time while going about their daily commute due to their small size and ability to move on narrow streets. An important component in a motorcycle is the motorcycle frame, the motorcycle frame is a useful part to support the weight of these components in the motorcycle vehicle system. However, it is rumored that a motorcycle frame with the E-SAF type has poor quality, so a sentiment analysis is needed. This research aims to collect the number of comments, both positive and negative, from social media users X about the E-SAF framework, and also to determine the accuracy of the application of the Naive Bayes method. The datasets collected from social media X amounted to 756 datasets. Then after going through the stages of data cleaning such as cleansing, tokenize, and stopword filters, the data that can be used for this research amounted to 696 datasets. The next stage is data labeling, namely by dividing the dataset with a ratio of 60:40, namely 60% of the training data totaling 417 datasets that have been manually labeled with the results of 224 negatively charged data, 193 positively charged data while the test data is 40% with a total of 279 datasets which will later be automatically labeled with the implementation of the Naive Bayes method. The next stage is that the test data goes through the data processing stage so that the test data is ready to be implemented into the Naive Bayes method. After implementing the Naive Bayes method, the accuracy obtained was 70.27% with a precision of 76% and also a recall of 79.17%. There was also a true Positive data of 57 and a true Negative data of 21. Data Visualization also displays words that appear frequently in the dataset. Here it shows that the Naive Bayes method is quite effective for the classification of sentiment analysis