Sentiment Analysis on Amazon Electronics Product Reviews using Machine Learning Techniques

Sentimental Analysis is one of the major functions of Natural Language Processing (NLP). It has gained much attention and importance in recent years since the great boom of data. The Internet is an unstructured and rich source of information that contains a large number of text documents offering opinions and reviews. Individual decision-makers, businesses, and governments can all benefit from the recognition of sentiment. In this research paper, we apply Sentimental Analysis on Amazon Products Review Dataset and aim to compare the accuracies of various Machine Learning algorithms, to find the most accurate one. Data used in this study is Amazon Products Review Dataset which contains thousands of reviews of various products sold on Amazon.com. The Dataset consists of 20,000 product reviews, with a balanced mix of positive and negative reviews and having a word limit of 150 for each review, for more precision. We have employed the Bag-of-Words (BoW) and Term Frequency - Inverse Document Frequency (TF-IDF), these machine learning feature extraction techniques. The pertinent features are then fed into six classification methods (Naive Bayes, Logistic Regression, Decision Tree, K-Nearest Neighbor, Random-Forest, and Kernel Support Vector Machine) and two deep learning models (Artificial Neural Network and Recurrent Neural Network).