Sentiment analysis is a machine learning tool that analyzes texts for polarity, from positive to negative. By training the machine learning models with examples of emotions in text, machines automatically learn to detect sentiment without human input. Machine learning allows computers to learn new tasks without being expressly programmed to perform them. Sentiment analysis models may be trained to read beyond mere definitions, to grasp things like, context, sarcasm, and misapplied words. This paper proposes a Deep Learning based approach mixed with Transfer Learning to avoid wasting resources. We also compare different Machine Learning Models for accuracy and iterations. We achieved an F1 score of 0.6 with LSTM, 0.5 with SVM and 0.7 with Vader. So, we conclude that using transfer learning is way more resource-efficient than training a model from scratch.