-
Notifications
You must be signed in to change notification settings - Fork 255
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add Wilcoxon's signed rank test #237
Conversation
comparisons/wilcoxon/README.md
Outdated
|
||
```python | ||
wilcoxon = evaluate.load("wilcoxon") | ||
results = wilcoxon.compute(predictions1=[-7, 123, 43, 4, 5], predictions2=[1337, -9, 1, 2, 3]) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hrmm should probably make some of these floats to make clear that those are allowed too
The documentation is not available anymore as the PR was closed or merged. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hi @douwekiela, thanks for the super clean PR. Left a comment regarding the feature types related to your own comment.
comparisons/wilcoxon/wilcoxon.py
Outdated
features=datasets.Features( | ||
{ | ||
"predictions1": datasets.Value("int64"), | ||
"predictions2": datasets.Value("int64"), | ||
} | ||
), |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If you want it to work for both int and float you can pass a list of dataset.Features
and it then automatically detects which one works. You can have a look at BLEU. Alternatively, I think floats would probably work both cases anyway, no?
Add Wilcoxon's signed rank test for comparing model predictions, e.g. for testing whether the difference in BLEU score between two models is significant.
Add Wilcoxon's signed rank test for comparing model predictions, e.g. for testing whether the difference in BLEU score between two models is significant.
Figured it'd be good to add a few more comparisons