Pengukuran Kinerja Optimasi Algoritma Bat Pada Algoritma Naive Bayes, KNN Dan Decision Tree Untuk Sentimen Analisis Di Lini Masa Twitter

Abstract

Social media is a very effective communication media in today's digital era. Of the social media, Twitter is the most widely used social media. Many tweets entered on Twitter have encouraged research in the field of text mining. One of the branches of text mining is sentiment analysis. Sentiment analysis in this study was formed from 3 classification algorithms, namely Naïve Bayes and Decission Tree. In practice, the results of the 3 classification algorithms often produce very low levels of accuracy. Bat algorithm is an algorithm that can optimize the results from the accuracy of the Naïve Bayes, K-NN algorithm and Decission Tree. In this study, two research scenarios were made: first, calculating the accuracy of the Naïve Bayes, K-NN algorithm, and Decission Tree. Second, optimizing the classification results of the 3 algorithms with the Bat algorithm method, which then re-tested the accuracy value. In the first scenario the percentage is generated from the accuracy value of Naïve Bayes of 33,58, K-NN of 33,61 and Decission Tree of 32,82. In the second scenario, using one of the objective functions, namely f(x) = x2, the Naïve Bayes value is obtained 39,01, K-NN 66,15 and Decission Tree 76,63. From the results of 3 the optimization test of classification Algorithm, it was found that the overall objective functions of the Bat algorithm were all able to increase the data accuracy value from before optimization. From all the tests, it was found that the Decision Tree algorithm has the highest average value of optimization increment, namely 43,81 %