Analisis Sentiment Pelanggan Terhadap Penilaian Produk Pada Toko Online Shop Amreta Menggunakan Metode Naïve Bayes Classification

Abstract

Sentiment analysis or opinion mining is an analysis that aims to see the sentiment of people or groups regarding certain entities. The sentiments expressed by society can be in positive, negative and neutral form. One media that can be given an opinion by the public is in the e-commerce application, namely the shopee application, shopee has a comment or assessment feature on the product that has been purchased. Toko which was used as a sample of researchersis an amreta online shop store , based on the results of the identification of the problem, it was found that the fact was that many comments did not match the stars given so it can be said that the rating cannot represent that the store's performance is good or not. Therefore, to increase the profit of shop work, the amreta still needs to evaluate the store. In conducting an evaluation, the store needs to classify positive, negative or neutral comments. Analysis of customer sentiment towards product assessments in amreta online shop stores using the naive bayes classification method. The use of test data in this study was obtained from the sentiment of amreta online shop consumers as much as 2014 data,then the data was processed through the data cleaning process resulting in net data of 1899 data. Furthermore, the data preprocessing process is divided into 3 stages, namely Tokenize Data, Transform case and Stopword removal. After that, the analysis of data for the automatic labeling stage using Text Vectorize from the process obtained data division into 3 data groups of 71% or 1343 positive data, 3% or 52 negative data and 26% or 504 neutral data. furthermore, it is processed using rapidminer tools while for operators in the form of algorithms using the Sentiment Naïve Bayes Classification model through automatic calculations. The results of the study can be concluded that the test data obtained have an accuracy level of 97.16% using the Naive Bayes Classification model.

The Shopee Application User Reviews Sentiment Analysis Employing Naïve Bayes Algorithm

“…Muhammad Muslimin, Tri Herdiawan Apandi, and Didik Garbian Nugroho presented studies focusing on Naïve Bayes classification, attaining accuracy rates of 94.38%, 91.33%, and 80.00%, respectively [2]. A study by Muhammad Muslimin et al in 2023, titled "sentiment Analysis of Rising Basic Commodity Prices Using Naïve Bayes Classifier," concluded that sentiment analysis on increasing basic commodity prices yielded negative responses, achieving an accuracy of 94.38% [1]. Didik Garbian Nugroho's 2016 study, "sentiment Analysis on Online Motorcycle Taxi Services Using Naïve Bayes," achieved an 80% accuracy rate utilizing sentiment data from Twitter [2].…”

“…For instance, Muhammad Muslimin achieved a Naïve Bayes accuracy rate of 94.38%, while Tri Herdiawan Apandi achieved 91.33%, and Didik Garbian Nugroho achieved 80.00% [1]. In a similar study, Muhammad Muslimin et al (2023) conducted research titled "Sentiment Analysis of Rising Basic Commodity Prices Using Naïve Bayes Classifier," concluding that Sentiment analysis of rising basic commodity prices garnered negative responses from the public. The Naïve Bayes method used in the analysis achieved an accuracy of 94.38% [1].…”

“…In a similar study, Muhammad Muslimin et al (2023) conducted research titled "Sentiment Analysis of Rising Basic Commodity Prices Using Naïve Bayes Classifier," concluding that Sentiment analysis of rising basic commodity prices garnered negative responses from the public. The Naïve Bayes method used in the analysis achieved an accuracy of 94.38% [1]. Another study by Didik Garbian Nugroho et al (2016) titled "Sentiment Analysis on Online Motorcycle Taxi Services Using Naïve Bayes" achieved an 80% accuracy rate using Sentiment data from Twitter, consisting of 3000 sentences categorized into 1000 neutral, 1000 positive, and 1000 negative sentences [2].…”