Algoritma Naïve Bayes Classifier Untuk Analisis Sentiment Pengguna Twitter Terhadap Provider By.u

Abstract

The development of the internet which has increased in recent years has made it easy for people to give their opinion on a product. By.u, as a new internet service provider, has made many new users share their opinions with each other. Many by.u users give their opinions through social media, especially twitter. From these problems, research was conducted using sentiment analysis. The research stages consisted of collecting data from social media Twitter, preprocessing data, weighting TF-IDF data and classifying using the Naïve Bayes Classifier algorithm. To get the best evaluation results, a comparison of training data and test data was carried out. Data classification is done automatically after cleaning the data in the preprocessing process. There are 2 labels for the data resulting from the automatic classification, namely positive and negative. The dataset after classification will be used as training data and test data. The datasets to be tested are divided into 3 numbers, namely the number of 1000 datasets, 2000 datasets, and 3000 datasets. The test was carried out 3 times for each dataset. The accuracy test is carried out using a confusion matrix. The test results with the highest accuracy were obtained by the nave Bayes classifier with a multinomial model of 85%.