Implementation of K-Nearest Neighbor (KNN) and Logistic Regression Algorithms in Sentiment Analysis Evermos App Reviews

Abstract

The Evermos application is one of the e-commerce applications for resellers and dropshippers in Indonesia. To find out the quality of the application requires an assessment from application users. One method that can be used is sentiment analysis. Sentiment analysis is one of the text mining techniques that can help in processing a judgment, complaint, perception or response to a particular object. In this study, the sentiment analysis carried out was the analysis of Evermos application review text on the Google Play Store accompanied by emoji conversion by comparing the K-Nearest Neighbor (KNN) model and Logistic Regression. The comparison of these two classification models was done to get the best accuracy in the sentiment analysis. The results of the two test models that have been carried out, the Logistic Regression model is a classification model that has the best accuracy results with a value of 98.1% in test data and 82.7% in training data.