Can we export the statmodel report to any external node or just have copy button here in the console because for some reason academicians needs to copy paste the statistics as it is without any changes. Do knime does not want that is needs be cited as much as papers. Then I think knime is missing “Word of mouth” impact. At least a report generator for Knime console output that should act like a journal.
@knime200087 what you can do is capture the output into a text file and store that. This code sample is from this workflow, assuming you have a path variable (string) where to store the file (var_path_validate) and a model name (var_model_name_full):
var_txt_summary = knio.flow_variables['var_path_validate'] + var_model_name_full + ".txt"
# bring the path of the summary back to KNIME as Flow Variable
knio.flow_variables['var_txt_summary'] = var_txt_summary
# capture the model summary in an TXT file
import sys
sys.stdout = open(var_txt_summary, 'w')
# do whatever it is that should be captured
print(extracted_model)
sys.stdout.close()
4 Likes
Thanks a lot, I just discovered yours way using GPT4 prompting, Also there is another problem with that though I can save it into doc file too. The problem is it doesn’t give you full split table in which all values are in separate cell. Rather I still have to copy that portion to excel and has to use cell split in excel to make it full formatted for publication quality.
But still I in this ultra modern age, who want to do lots of steps. I think knime developers smart enough to enable the embedded internal “Jupiter like” features (It is almost done) like the plots are now can be visualize in the preview panel etc.
But still I think it is not enough. Because there is lots of “Disruption” is going out there. And knime is just in “almost there” place. But overall I am happy till to date.
For some models you use in a Python environment you could extract specific values into KNIME Flow Variables and then bring them to KNIME.
My impression is sometimes it also depends on the developers of packages to provide and document the extraction of parameters to be used in another way then display in a notebook. I often have to find ‘creative’ ways to extract such information and store them for further use and reference. The good news is KNIME as a platform can handle that although it might not always be pretty.
The idea to sort of fully integrate Jupyter into the KNIME node might be interesting. The option to directly export the results.
Can you elaborate? Or maybe start another thread? For GPT4All you can also export to HTML. Not sure if this then is ‘Publication’ quality. Another option might be to explore the new Reporting feature and include a Table view in a component and then export that as a PDF, but I have not tried that.
1 Like
Again thank you for your effort, but the problem is my report generated in console is multiple table and multiple analysis in one single console with diffrent table structure, that is very special case, As per my knowledge “Save as Doc or Text” can have some solution for that, because in that case I can copy that to excel and split them table by table.
However , in that case developers can create a spss like output, for multiple table report from one console.
1 Like
@knime200087 from your screenshot you seem to use ols. A Google search brings this suggestion that results are actually stored as pandas dataframes which could easily brought back to KNIME or be stored as Parquet files.
Or so it seems
Maybe you can provide an example. That then maybe could be refined with ChatGPT.
3 Likes