An Exploration of Sentiment Analysis Techniques Enhancing Customer Purchasing Behavior Prediction of Smartphone using Machine Learning Classifiers for Improved Performance

Abstract

This research investigates the impact of machine learning classifiers on smart phone product reviews.  Every action taken by a living being carries feelings because the world is made of sentiments and emotions. Opinion mining is another name for Sentiment Analysis, and it is a classification process whereby machine-learning techniques are applied to text-driven datasets to analyze the emotion or opinion expressed in a text. The sentiment analysis technique extremely helps many organizations to quantify customer fulfillment with specific products based on reviews of customers in a very faster way.  In this study, a sentiment analysis model is proposed for categorizing product reviews into negative, neutral, and positive ones. But in the existing system, the neutral review is not considered which will result in an error in the customer’s opinion. In the existing project, various algorithms like Decision Tree, Naive Bayes-NB, Support Vector Machine-SVM, Maximum Entropy and k-Nearest Neighbor-KNN are used to categorize the accuracy of the result. Among these Maximum Entropy and Naive Bayes performs well and produced better accuracy. The proposed research uses the Random Forest methodology for the training stage and the testing stage. A dataset of 50,000 reviews was used for this research concerning Smartphone products. While compared with other machine learning models Random Forest machine learning technique greatly improved the accuracy.