This is an NLP and Flask based application which involves predicting the sentiments of the sentences as positive or negative. The classifier is trained on a huge dataset of IMDB movies reviews. The model is then hosted using Flask to be used by end users.
This project has text pre-processing done through NLTK and Regex and EDA for understanding the features and data well. The text is then coverted into vectors using 2 techniques - Countvectorize and TF-IDF. Two Machine Learning algorithms (Naive Bayes and SVM) are then used with combonitions of above 2 techniques and it is found that Naive Bayes with TF-IDF.
The model is then saved in a Pickle file and used in the Flask Application to host the website on localhost.