Comparison on Naive Bayes and K-Nearest Neighbor for Online Transportation using Sentiment Analysis in Social Media
Aldy Rialdy Atmadja, Wisnu Uriawan, Ferdinand Pritisen, Dian Saadillah Maylawati
Department of Informatics, UIN Sunan Gunung Djati Bandung, Jalan A H Nasution 105 Bandung 40614, Indonesia
Abstract
Nowadays, online transportation is one of the transportation that is increasingly preferred by people. It becomes important because people need transportation to be more effective and efficient. However, sentiment analysis is necessary to improve the quality of services on online transportation. Sentiment analysis includes the process of extracting opinions, sentiments, evaluations, and emotions of people about online transportation services on Twitter social media. To get more accuracy in classification, the opinion is taken in large amounts and classify into positive and negative class. There are several steps that use sentiment analysis. Data collection, preprocessing data, POS Tagging, and opinion classification use the Naive Bayes Classifier method, compared to the accuracy of the K-Nearest Neighbor method. The results of the comparison of Naive Bayes Classifier and K-Nearest Neighbor algorithms uses 565 data tweets from Twitter, divided 500 trained data, and 65 test data. The result showed that the Naive Bayes Classifier algorithm had achieved the accuracy rate of 66.15%, and K-Nearest Neighbor algorithm produces the accuracy rate of 67.69%. From the results, the K-Nearest Neighbor algorithm perform better accuracy in sentiment classification than Naive Bayes Classifier.
Keywords: Online transportation, Sentiment analysis, Twitter, Naive Bayes Classifier, K-Nearest Neighbor
Topic: Computer Science