Comparison analysis of Naive Bayes Classifier (NBC) and K-Nearest Neighbor (KNN) methods on user reviews of the Google Maps application

Abstrak

Data is a collection of facts, where these facts can provide an overview of a situation. Data can be stored in various ways, for example, data about applications that are stored in a database server and have various types such as text, image numbers, and others. Google Maps is a free service from Google with functions like world maps which can be accessed using a browser or application. On the Google Play Store page, there are reviews and information about a product or application that is stored in the form of text, score, or something else. This research conducts sentiment analysis on user reviews of the Google Maps application in the Google Play Store. Sentiment analysis is carried out as a tool for classifying or categorizing information in the form of text into positive and negative categories or labels so that application developers can find out the advantages and disadvantages of their applications. The process for carrying out sentiment analysis is like doing text preprocessing and word weighting which aims to give value or weight to the words contained in a document. Then the classification method used in this research is the Naïve Bayes Classifier and K-Nearest Neighbor, then it will be visualized with a word cloud. The accuracy for the Naïve Bayes Classifier is 80.6%, while for the K-Nearest Neighbor it is 78.8%. Based on these results indicate that the Naïve Bayes Classifier method is better in classifying. Meanwhile, in visualizing with the word cloud, words that have negative labels, such as "point", "please", "accuracy", "location", and so on are obtained. Then for those with positive labels, they include "helpful", "good", "okay", "accurate", and so on.