I have finally come up with a recommender system for the supermarket which now suggests products to users based on implicit collaborative filtering. But I am stuck at a point where I do not know how to proceed any further as it needs to be validated i.e, how good this is. Any ideas and suggestions are welcome.
One way to go is to hide some data and train the system with the rest to see if it recommends the products that were hidden.
However this approach won't be be of much use for costumers that haven't bought anything yet.
Another way is for you to manually evaluate some of the recommendations made by the system.