This project focuses on analyzing sentiment in a dataset of 25,000 unique IMDB movie reviews, classifying them into Negative, Neutral, or Positive categories. The analysis involves deep understanding of the context, semantics, and sentiment expressed in the reviews. Advanced machine learning models including Simple RNN, GRU, and Bi-directional LSTM were developed and implemented to achieve this classification. By employing natural language processing techniques such as tokenization, lemmatization, stemming, punctuation removal, and stopwords elimination, the models reached an impressive overall accuracy of 92%. These results provide users with quick and reliable insights into the sentiment trends of movie reviews. The project not only showcased the effectiveness of different RNN architectures in handling text data but also highlighted their capabilities in sentiment analysis.