Basically Machine Learning workbench is not only about machine learning because of the ability to use Python.

A) General use cases

-GIT integration

-Scheduled transfer of the configuration between teams (release management)

-Checking / transforming data that's not possible with Vertica SQL or PQL

B) ML use cases

-Fraud Detection

-Predicting Behaviour

-Transportation Optimization

-Process Automation

I would focused on easy ML use-cases for python and then watch out for the maximum memory limitation (default: 5gb, it can be upgraded by Celonis if needed). You can get the data with PyCelonis + Pandas.

Some ready-to go examples were also avaiable on Celonis Academy:

https://academy.celonis.com/learn/course/machine-learning-duplicate-checker/machine-learning-duplicate-checker

Best Regards,

Mateusz Dudek