Repository navigation
Replies: 4 comments 4 replies
|
It gets even weirder when I count missing values via Python Script and pandas (using pd_df.isna().sum()). Shows 0 missing values for all meta attributes. Yet, Orange's Data Table widget correctly finds % of missing values in the meta attributes. Did this in Colab and ther are definitely missing values there but Python Script does not see them, although Data Table does! |
|
So what you are seeing is probably connected to the fact that empty strings count as Unknowns in Orange. I understand that could be a bit cumbersome, but this limit comes from files readers. For example, if you have an Excel file you don't really have a separate "unknown" value. Thus, empty strings count as unknowns. |
|
Understand that using, e.g. fillna('') would still count as unknown. However, the problem is that Orange does not see missing values from a Python Script but it does via an Orange table its missing values stats! |
|
Tried to send you an Orange workflow file to show you this but GitHub does not allow this file type. |
Uh oh!
There was an error while loading. Please reload this page.
Hi,
I am working on an NLP dataset, which contains a number of meta attributes where some have missing values.
Can you please tell me how to fix this issue? I've been trying but there does not seem to be a way to handle missing values for such attributes in Orange.
Even when I try to fix the issue via Python Script widget and pandas fillna('') or fillna('Unknown'), Orange still says there are missing values.
Is this a glitch or I am doing something wrong?
Many thanks for your help!
All reactions