A Comprehensive Exploration of Stack Ensembling Techniques for Amazon Product Review Sentiment Analysis

This review paper delves into the realm of sentiment analysis in e-commerce, specifically focusing on Amazon product reviews. Leveraging a stack ensemble machine learning model, we explore the intricacies of sentiment understanding and compare its performance against established models like Naive Bayes and LSTM-based approaches. The methodology involves meticulous data collection from Kaggle, preprocessing through text cleaning, tokenization, and lemmatization, and the construction of a stack ensemble model incorporating support vector machines, random forests and decision trees. Model performance has been evaluated using a variety of metrics, including confusion matrix, F1- score, recall, precision, and accuracy. Our comparative analysis reveals nuanced insights into the proposed model's strengths and weaknesses, showcasing its potential for advancing sentiment analysis in the ever-evolving landscape of e-commerce.