Hi.

Machine Learning workbench is just a linux virtual machine with specific limitations on installed packages, memory, disc space and not having root access. Because of that usually it will limit the usage mainly to data manipulations and interconnectivity using PyCelonis, Pandas, Requests, Matplot lib, NumPy and other python packages. Event if you cannot install new linux packages, you can still install new python packages, so that's your biggest limitation.

For me it was mainly used for:

A) Custom python extractor for 3-rd party API services (Hate extractor builder)

B) Custom data manipulation: f.e. using machine learning algorithm like K-means clustering

C) Custom data manipulation: generating charts and maps that cannot be generated in Celonis itself using f.e. Matplotlib for charts and Folium & OpenRouteServices (https://openrouteservice.org/example-avoid-flooded-areas-with-ors/)

D) Workarounds for missing PyCelonis capabilities like listing current celonis users (Example can be found in Celopeers site, I've posted code snippets along with Gulermo)

Best Regards,

Mateusz Dudek

Hi Mateusz ,

Glad to see you again. Thank you for your update regarding this topic can you please suggest any documentation to refer for your second point (custom data manipulation).

Best Regards,

Akshay

Hi Akshay,

regarding second point: you can search some guides related to Machine Learning or other things that you cannot do by default in Celonis (f.e. image recognition using tesseract) - please note that VM capabilities are limited, so I would be careful about using it for large calculations or generating a lot of data. Basically that's more a python topic than celonis topic

Example of usage for ML algorithm in Python: https://realpython.com/k-means-clustering-python/

Best Regards,

Mateusz Dudek

Hi Mateusz,

Thank you for providing more insights on this topic.

Best regards,

Akshay