Is your feature request related to a problem? Please describe
Currently we are using simple linear regression and power law regression to compute the scaling of resource consumption of pipeline steps (duration, cpu consumption, memory usage). There might be ML-approaches that could perform better.
Describe the solution you'd like