Christ-University-Annual-Datathon-1st-place-approach conducted by Mahindra Comviva

This is the solution to the Annual Christ University Datathon. I was placed 1st in the hackathon.

Solution details:

The dataset was provided by Mahindra Comviva. The dataset mainly consisted of tweets relating to Airtel (One of Mahindra Comvivas biggest clients).

1. The main model was a GRU-LSTM-Capsule Model for twitter sentiment classification.
2. Used Cyclic Learning Rate which improved score considerably.
3. Achieved 89% accuracy.