SMS Classification using NLP and Bag of Words

Aim was to classify SMS with financial transaction data. I built bunch of NLP and Bag of Words based classifiers using methods such as Count Vectorizer, TF-IDF Vectorizer, Word-2-Vec Vectorizer, Logistic Regression, Naive Bayes and XGBoost. The final model was an ensemble which gave Accuracy of more than 98%. This model, in production serves more than 200 Mn SMS everyday and saves more than 36 daily man hours.