Evaluation of Feature Selections on Movie Reviews Sentiment
Danny Oka Ratmana, Guruh Fajar Shidik, Ahmad Zainul Fanani, Muljono Muljono, Ricardus Anggi Pramunendar · 2020 International Seminar on Application for Technology of Information and Communication (iSemantic) · 2020
In the Text classification task, feature selections are one of the methods to improve classifier performance. With dimension reduction of the original features, it usually used to get better performance of accuracy, precision, recall, or maybe to accelerate computation time. In this paper, we applied several feature selections method such as Kbest with Chi-Squared Selection, Linear SVC, and Tree-based Selection into five classifiers: Naive Bayes (NB), Decision Tree (DT), K-Nearest Neighbor (KNN), Support Vector Machines (SVM) dan Neural Network (NN). Datasets that we used are collected from Kaggle, Imdb Movie Review 5000 records, and the best F1-Score results are on Linear SVC that running on SVM Classifier 92,32%.