Brand Review Prediction using User Sentiments: Machine Learning Algorithm

As the innovation is increasing exponentially, the manner in which individual expresses their opinions in several website is also changing. There are plenty of reviews and opinions on platforms like Facebook, Twitter, Yelp. Typically scores are on a range of 1–5 levels or stars. Identification of sentiments by investigating textual data has played a crucial role in studies on analytics as it provides valuable alternatives for sentiment mining. A review is an analysis by an individual about items or service that has acquired using the service or item, or had expertise with it. Any ecommerce site's ranking is strongly based on its user's perception. This paper aims to report operation of Machine Learning (ML) algorithm on Yelp's database to evaluate, anticipate, and suggest brand. We implemented Naive Bayes, Random Forest, Decision Tree, Support Vector Machines, K- Nearest Neighbor and Multilayer Perceptron classifiers with sentimental analysis algorithm. Multilayer perceptron classifier achieves highest accuracy of 93.40 %.