This article explores sentiment analysis through reviews of Amazon's Alexa products, which are analyzed in a specific dataset. A review of Amazon's Alexa products’ dataset and analysis of its features took place during our examination. It explores the evaluation process of machine learning analysis on the dataset and its mechanism for positive and negative review classification. The modeling stage of CRISP-DM in this experiment uses the classification approach that features Random Forest together with Decision Forest, XGBoost model, and K-Nearest Neighbors (KNNs) as classification algorithms. The Random Forest algorithm stands out as the most accurate model since it reaches 93.28% accuracy, followed by the XGBoost, which demonstrates 92.80% accuracy. Next is the Decision Forest, which stands at 90.25%, and the KNN algorithm stands as the least effective model in this comparison because it reaches an accuracy of 80.12%.