Sentiment Analysis of Amazon Product Reviews

Abstract

In this study, we conduct the process of sentiment analysis to reviews of the Amazon Alexa product through the use of machine learning models: XgBoost, Random Forest, and Decision Tree. The data set is taken from Kaggle and consists of customer reviews that are either positive or negative. In the first stage of pre-processing, we clean the text, tokenize the text, remove stop words and stem the text of the data. Next, we split the data set into training and testing data set for the model to learn and classify sentiment. The results of the experiments showed that XgBoost returns higher accuracy and generalization than the other models. Therefore, we chose XgBoost as the model of choice for sentiment classification and is helpful for informing business if their customers are satisfied or dissatisfied.