Closed vidhisshah closed 4 years ago
Thanks, @vidhisshah . Will check this evening and reply back.
@vidhisshah ,
Thanks for your detailed report. Much appreciated.
FairModel has an attribute called _tree
. This attribute is a FairDependencyTree instance that is used to track calculation dependencies. This class was using erroneously using "Contact" and "Action" as handles for tracking, which is where the issue arose.
I am going to take this fix and another I've been meaning to get out (#24) and package them up this weekend for version 0.1-alpha.11.
For posterity. going forward these will work:
These will not work:
@vidhisshah ,
Would you please download version 0.1-alpha.11 from PyPI, ensure it fixes your problem, and then close this issue out?
Thanks again for the bug report.
@theonaunheim Thanks a lot!
The keywords 'C'
,'Contact Frequency'
,'A'
, and 'Probability of Action'
are working fine.
One small issue though, it doesn't seem to be doing the check where values need to be between zero and one. Could you please check that? For example, model.input_data('Probability of Action', low=2, mode=10, high=20)
would have thrown an error earlier, but it seems to be accepting it now.
Brilliant, and thanks for your attention to detail, @vidhisshah !
It appears that FairDataInput
validator and accompanying unit tests use the term "Action" as well. I will modify FairDataInput._le_1_targets
attribute, look for any stray "Action" references in the code base, and check back with you in the next day.
@vidhisshah , I believe I have fixed the validation problems.
At your convenience, would you please check version 1.0-alpha.12 on your machine, and close out this issue if fixed?
Thanks again for your diligence.
Works! Thank you.
These inputs throw a key error:
This works fine:
model.input_data('Contact', low=10, mode=30, high=40 )
Similarly,
'A', 'Probability of Action'
fail but'Action'
works.I've tested all other abbreviations from target_map and they seem to work fine. Please do check.