Sentiment analysis is part of text mining, which means extracting data from comments on specific products and people’s opinions or sentiments. It involves considering opinions, analysis, and suggestions to customers about Amazon Alexa products by classifying a review as positive, negative, or neutral. This work is to explore different learning algorithms to solve sentiment analysis on Amazon products. The proposed work is a comparative analysis of existing learning algorithms that can help businesses in real-time decision-making by providing insights into customer preferences, needs, and choices which in turn results in the growth of the company and enhancement of customer satisfaction. The existing works provided the accuracy of classification to a level of 80%. Algorithms used in previous projects are SVM, Naïve Bayes, Random Forest, and other learning algorithms. In this work, logistic regression, Gradient boosting, Multinominal NB, and Gaussian NB are used along with previously implemented algorithms for sentiment analysis. Deep learning algorithms such as Long Short-Term Memory (LSTM) and ANN with dropout and without dropout layers are also implemented and compared with machine learning algorithms. Word2Vec is employed with LSTM to increase the accuracy of the representation. Based on the evaluation of their metrics, accuracies are compared.