Is there a ML technique applicable to this problem? (knapsack variation with an item of 'variable' volume)

I'm very, very new to machine learning, but I'm dealing with a problem that seems like it might be a candidate for a ML approach. It's basically a knapsack problem, with a twist. I've got a set of items (A, B, C,...) each with a known volume (the percentage of the container it will take up). However, there's another item, X, which often fits in the 'spaces between' other items, but how many X's a container can hold depends on what other items are in the container. For example, if the container holds 2 A's and 1 B, then I can fit 4 X's. However, if I add a C to the container, I can now only fit 2 X's. I have some historical data to work with, so I know in the past the various configurations and how many X's they were able to contain. Is there a way to construct a model based on this historical data so that I can feed it a proposed consolidation, and get back how many X's such a configuration could contain?

Any suggestions would be appreciated (even if it's just to tell me I'm on the wrong track and should look into some other-- non-ML-- approach). Thanks!