I want to import a manually transformed table from datapool to ML workbench to do some operations. Is it possible to do that without loading the data model? Can i import the transformed table directly?

you can, but it will download only 100 records.

Create a temporary process model with just the table you want to download, then use PQLquery to retrieve all the data into a dataframe.

Then delete the temporary process model

HTH

is there any proper documentation of how to do this?? if possible can you share a notebook?

i am using pycelonis version 1.7.4

Sure.

Add data (and table) to a datamodel

https://celonis.github.io/pycelonis/1.7.4/tutorials/api_tutorial/02_Adding_new_data_to_data_model/

Get the data from a table in a datamodel

https://celonis.github.io/pycelonis/1.7.4/tutorials/api_tutorial/01_Pulling_data_from_dm/

To create the datamodel you need to use

pool=celonis.pools.find(poolname)

HTH

dm=pool.create_datamodel(data_model)