Hello All,
I have a question, if we do not have enough records to train the PI then can we use the live cases (which are closed) to train on the data that we are getting and get to a point where we can test the predictions are satisfactory enough and match the threshold and confidence so that we can push it to real-time usage?
(or)
Would you suggest that once there are at least 10k cases in the PROD, cloning it down to the lower instances (DEV) and training the model there and keep training it until required results are derived and then move to PROD?
