I have to my entire disposition an enormous amount of data related to grocery ecommerce, mostly about when and what products have been bought by the customers and some data related to their personal profiles (state, type of id document, age, etc. although not all of these fields are complete).
In a recent past, I have contributed to build a recommendation system using the libfm library but that project has finished and I want to invest my time in another project.
I have tried to make a model to predict what are the product each client will buy by using a multi-label framework but each client usually buy more than 20 items making the model very unstable (or I was not able to build a good model ).
Do you guys have some ideas to begin a project using this sort of ecommerce data?
Thanks in advance!!