Understanding H2O AutoML event logging
Since H2O AutoML automates most of the ML process, we have given some control to the machine. Encapsulation means that all the complexities that lie in AutoML are all hidden away, and we are just aware of the inputs and whatever output H2O AutoML gives us. If there is any issue in H2O AutoML and it gives us models that don’t make sense or are not expected, then we will need to dig deeper into how AutoML trained the models. Hence, we need a way to keep track of what’s happening internally in H2O AutoML and whether it is training models as expected or not.
When building such software systems that are aimed to be used in production, you will always need a logging system to log information. The virtual nature of software makes it difficult for users to keep track of what is going on as the system does its processing and other activities. Any failures or issues can lead to a cascade of underlying problems that developers may end up...
