PNR Prediction, Powered by TensorFlow
When we decided to add PNR status prediction to Indian Train Status, we wanted it to be state of the art. This meant Machine Learning, and TensorFlow (open-source machine learning framework by Google) was the obvious choice.
We proceeded to train a deep neural network classifier to predict the final status based on several parameters including the train, route, class, present status, and day of travel. TensorFlow provided us with powerful tools like multi-dimensional embeddings, adaptive learning and visualisation which helped to extract intricate patterns from the complex data.
On validation, we have achieved an accuracy of ~84%. We believe this feature will be as good or better than any other PNR prediction service in the country.
So next time you get waitlisted, fight back by checking your chance of confirmation in ITS - its state of the art!
Get ITS app now•••
We proceeded to train a deep neural network classifier to predict the final status based on several parameters including the train, route, class, present status, and day of travel. TensorFlow provided us with powerful tools like multi-dimensional embeddings, adaptive learning and visualisation which helped to extract intricate patterns from the complex data.
On validation, we have achieved an accuracy of ~84%. We believe this feature will be as good or better than any other PNR prediction service in the country.
So next time you get waitlisted, fight back by checking your chance of confirmation in ITS - its state of the art!
Get ITS app now•••
