Let's say that a user (CostumerID) have only one transaction in the test set and no history in the training , how can we predict a fraud detection in this case ? any ideas !
I guess that is our job. I.e Being able to predict an entirely new customer without prior info.
I am asking about some useful ideas and ways to try
their are a number of ways to do that one is do supervised learning we have alot of materials to learn from online