Sentiment Analysis of User Reviews Based On Naïve Bayes Classifier Algorithm with Hyperparameter Optimization: A Case Study On Application "Kredit Pintar"

Abstract

The development of information and communication technology makes it easy for people to take advantage of their sophistication in various sectors, especially industry and the economy. However, due to the Covid-19 pandemic in Indonesia, it has an impact on the socio-economic community. To meet their needs, most people use online loan platforms because they feel they have easy requirements. One of the most trusted online loan platforms and protected by the OJK is "Kredit Pintar", which is the number one platform that is most widely used by the public to borrow money in 2021. From the facts, the purpose of this research is to find out the reviews of the community of online loan application users “Kredit Pintar” through sentiment analysis based on text data in the form of public comments taken from the Google Play Store site application of “Kredit Pintar”. The data used are 1374 reviews with positive and negative class classifications. This research was conducted by using an analytical text mining method using the Naïve Bayes Classifier algorithm and applying Hyperparameter Optimization by comparing two models of GaussianNB and MultinomialNB functions by Phyton programming. The results of the classification of training and testing data show that based on performance evaluation, the best model uses the MultinomialNB function with an accuracy of 97.08% for training data and 90.54% for testing data.