Yelp Review Classifier
Goal
To classify reviews into 1 or 5 stars based on the text content.
Process
Implemented NLP techniques to classify reviews into 1 or 5 stars based on the text content with 84% accuracy. 1. BOW: First, bag of words (technique of text modelling) was implemented via the CountVectorizer() function from sklearn. (lowercase -> remove punctuations and stop words: repeated words that contain very less info -> get frequency of occurrence and score them (vectors) 2. Tf-Idf ( term frequency-inverse document frequency): gets relevance of a word. 2 major components: TF and IDF. TF calculates by assigning equal weight to all of the words. IDF helps us get an idea of how important a word is. Formula: 3. Train: train on the Naives Bayes Classifier