Bug in Evaluation: save_path expects a "."

[randomn4me]

Hi,

the function evaluate_multiple_datasets expects the save_path to be splittable via a ..

Specifically the following line throws an error if this is not the case

ClustPy/clustpy/utils/evaluation.py

Line 459 in 625c7d5

save_path.split(".")[1])

I can create a PR but first I need to know what the code should do. Either it assumes a . (as is the case currently) then I would add it to the documentation and add an assert. Otherwise (which I think is the more generous approach) I would not assume the extension and just add it, i.e., with a given save_path = "intermediate_result" then I would change the inner_save_path to

inner_save_path = None if not save_intermediate_results else "{0}_{1}.csv".format(save_path.split(".")[0],
                                                                                  eval_data.name

And clarify the documentation to something like the string, that is prepended to the intermediate save paths (something like that).

In either way, I would add, that the save path should be a string multiple files are going to be generated prepended by that string. I suspected it to be a directory.

What do you think?

[collinleiber]

Hi,

good catch!

The best solution would be to re-think the general handling of paths in the whole package (see e.g. #77).
However, as a quick fix I would suggest to add the following code to evaluate_dataset and evaluate_multiple_datasets:

if save_path is not None and "." not in save_path:
        save_path = save_path + ".csv"
    assert save_path is None or len(
        save_path.split(".")) == 2, "save_path must only contain a single dot. E.g., NAME.csv"

This is probably the easiest way and is already included for the save_labels_path parameter.

Also you are absolutely right that the documentation has to be improved as it is totally unclear how the intermediate files are created.