In data science you sometimes encounter a scenario where you have meta data on a given process and the process data itself.
For example you have a mechanical component that is tested over time. So you have a (multi value-)timeseries (e.g. air pressure, temperature, mechanical movement, electric resistance and so on) on the one hand and the product metadata (age, CAD design parameters, used materials and so on) on the other hand.
In this scenario you can't really just use timeseries methods since you're ignoring the metadata, but just focusing on the metadata is also wrong, thus I'm looking for ways to use both types of data in my ML-system to better predict an anomaly or the likelihood of the test failing.
Do you know any models, tutorials or papers for this kind of data? What are some good starting points to get experience with this kind of data?