Import Error: epsilon_nondominated

[SnuggleBug91]

In the section on directed search the following import doesn't work.

'''
from ema_workbench.em_framework.optimization import epsilon_nondominated
'''

This is my first time writting an issue so not sure if you need any more than this.

[quaquel]

Thanks for reporting this issue. You have to be a bit more specific than this. I don't know what you mean with "the section on directed search". It would also help if you can indicate what version of the workbench you are using (print(ema_workbench.__version__)).

[SnuggleBug91]

Sorry, I wasn't particularly helpful with my comment. Apologies.

The error comes from following the tutorial on the ema workbench website: https://emaworkbench.readthedocs.io/en/latest/indepth_tutorial/directed-search.html

Specifically, when combing multiple runs of optimising by searching over levers then combining the non-dominated solutions into one dataset.

The code block in question is

from ema_workbench.em_framework.optimization import epsilon_nondominated, to_problem

problem = to_problem(model, searchover="levers")
epsilons = [0.05] * len(model.outcomes)
merged_archives = epsilon_nondominated(results, epsilons, problem)

The version of the workbench I'm running is 2.2.1

[EwoutH]

@SnuggleBug91 Could you update to 2.3.0 and see if the bug still persist? I'm not sure, but it might be fixed already in #150.

[omarcastrejon]

Hi,
Have you found a solution to this issue?
I am having the same problem trying to use epsilon_nondominated. I am using version 2.4.0 of the EMA workbench. Not sure what the problem could be but this is the error I get:

Any thoughts?
Thank you in advance.

[quaquel]

There is a fix on master, just need time to cherry-pick al bug fixes since last release to create a new release. You can still install master through pip (check google)

[omarcastrejon]

@quaquel thank you for your quick reply. I have installed the master version (2.5.0.dev0) but unfortunately I'm still getting an AttributeError and and EMAError (please see below). Any advice would be really appreciated. Thank you.

[omarcastrejon]

By looking at some documentation on DataFrame.itertuples, itertuples() does not work with column names that include special characters such as spaces, '-', '&', or '.'. My suggestion: either (1) create a note saying that the name of the levers should not include special characters; or (2) use named itertuples() by creating a temporary list of names to iterate over DataFrame rows as namedtuples.

[quaquel]

There is already an open issue for this (#274)