<![CDATA[Sentiment analysis is a process for extracting data in the form of textual, with the aim of obtaining information about the tendency to evaluate an object under study. Sentiments given by the general public can be used as a reference in making product decisions. Sentiment given can be in the form of positive, negative and neutral sentiments. One of the information technology products that has stolen enough attention in the last decade is Bitcoin. The purpose of this study is to compare several classification algorithms using Feature Selection. There are several classification algorithms that can be used for sentiment analysis, such as Deep Learning, Decission Tree, KNN, Naïve Bayes. Textual sentiment classification has constraints on datasets that have high dimensions. Feature Selection is a solution to reduce the dimensions of a dataset by reducing attributes that are less relevant. Feature Selection used is Information Gain and Chi Square. The method used to perform the comparison is by comparing the four classification algorithms to find the best algorithm, then comparing the Feature Selection to get the best between the two, then integrating the best classification algorithm and the best Feature Selection. The results showed that the best classification algorithm was Deep Learning with an accuracy value of 78.43% and a kappa of 0.625. The results of the comparison of Feature Selection, Information Gain get the best results with an average accuracy value of 63.79% and an average kappa of 0.382. The results of the integration of the best classification algorithm with the best Featrure Selection obtained an accuracy value of 78.63% and a kappa of 0.626 where the value was included in the Fair Classification category.]]>
Copyrights © 2022
