Classification and Sentiment Analysis of Amazon Alexa Reviews

This study is about developing sentiment analysis and classification method for Amazon Alexa products based on the rating and feedback given by customers. The purpose of the method is to investigate the polarity of positive or negative opinions expressed by the buyers of the products to attain insights and recommendations from the provided reviews by the customers. The Amazon Alexa Reviews dataset used in this work consists of 3150 reviews. It contains features related to the ratings, dates, variants, and customer comments on several Amazon Alexa products, such as Echo Dots and Alexa Echo. Four classification algorithms of Decision Tree (DT), Random Forest (RF), XgBoost (XB), and k-Nearest Neighbors (k-NN) are implemented, and the evaluation metrics of accuracy, precision, and recall compare their performances. The evaluation results of the four algorithms have revealed that the RF achieves the best performance among the four classifiers, with an average accuracy score of 94.17%, a precision score of 93.71%, and a recall score of 99.33 %.