Sentiment Analysis of Amazon Smartphone Reviews Using Machine Learning & Deep Learning

Abstract

The past few years have been marked by quite a few developments in e-commerce and online shopping with the biggest of them being in the smartphone segment. India is now the world’s largest market for smartphones with its share having increased to 45% in 2020 by registering a mammoth 7% growth during the pandemic year. Some of the major smartphone brands here are Xiaomi, Samsung, and OnePlus. These brands have often partnered exclusively with e-commerce platforms like Amazon and Flipkart with sweet deals and offers for buyers. For smartphones of all price segments, reviews on these sites can be an important indicator of how satisfied customers are with the product and can also be an important factor for decision making that helps customers choose whether a product is worth purchasing or not. In this paper, we will be exploring algorithms and techniques used for sentiment analysis and text classification of smartphone reviews on Amazon. The dataset we used for research is available on Kaggle and contains 6S,000 reviews of 720 smartphones of numerous brands. We have used a combination of machine learning and deep learning algorithms for the same, starting with baseline logistic regression and naive Bayes models and then moving on to complex support vector machines and Recurrent Neural Networks such as LSTM using the FastAI library.