SELEKSI FITUR INFORMATION GAIN DAN ALGORITMA NAÏVE BAYES UNTUK REVIEW OPINI KONSUMEN

Abstract

The growth of internet users in Indonesia is increasing, this is in line with online shopping habits or often referred to as e-commerce which continues to increase. Various things are done by e-commerce companies to maintain customer loyalty, one of which is through product evaluation using consumer opinion reviews. The number of reviews that are too many will be biased, so it is necessary to do a classification method that will help e-commerce companies to find out the extent of their customer loyalty. Consumer review becomes something important because all assessments of the products they buy are all in the review column. In this research, a consumer review is carried out using the Naive Bayes classification method and to improve the accuracy of attributes using the Information Gain feature selection and using the Select by Weight operator which will display the best attributes of the pre processing process. The review data set is taken from consumers' comments on Google Play. The results of this study are classifying consumer reviews into positive reviews and negative reviews with Cross Validation using 10 fold, the accuracy of the Naive Bayes method is 78.4% using the Information Gain feature selection method, the accuracy increases to 81.2%