Sentiment analysis utilizes natural language processing to obtain and classify sentiment from data in the text form. The paper emphases on the classification of mobile product reviews using various algorithms. The experimentation is conducted on the two datasets obtained after cleaning where one dataset accounts for the negations in sentences whereas the other dataset does not. The experimentation process involves combining the feature extraction techniques with the different algorithms. Feature extraction is implemented uses a Count Vectorizer (CV) and Term Frequency – Inverse Document Frequency (TF-IDF). The classification is executed utilizing machine learning techniques like Naïve Bayes, Support Vector Machine, Random Forest, Logistic Regression, and deep learning techniques, including Long Short-Term Memory (LSTM), Bidirectional Encoder Representations from Transformers (BERT) and a hybrid of BERT with Bi-Directional LSTM. The results show that among machine learning algorithms, RF performs the best with both extraction techniques giving accuracies of 0.9774 with TF-IDF and 0.972 with CV. Among deep learning models, the execution of BERT model by itself provides the best accuracy of 0.9863. It is observed that the dataset handling sentence negation improves the execution of all algorithms. This comprehensive evaluation highlights the effectiveness of both standard machine learning and advanced deep learning methods for analyzing sentiments.