-
-
Notifications
You must be signed in to change notification settings - Fork 403
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
Aggregation of measure timetrain is made on test set #1286
Comments
Agreed on both points. Could you make a PR for this please (including tests)? |
Yes, I will! Shall the test be in test_base_measures.R? |
We don't have I would put the tests in test_base_resample.R. |
We discussed this issue last week in our mlr meeting and @ja-thomas (ping :-)) had some arguments which argue against my idea of changing àggr |
Just some thoughts on why I think setting it to I would argue that in most cases we use So a "normal" thing to do is something like:
If we change the aggregation of |
please dont do anything here. i am not sure what the best solution is and have assigned myself. i also dont see this as too urgent. |
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. |
Hi,
while debugging #1284 I stumbled over this and this might be a bug:
If one wants to assess only the training time of a learner and uses a resampling strategy like 'cv' he/she will most likely choose to set
predict = "train"
in themakeResampleDesc
. But the the resampling won't work because the aggregation function is set to test.mean. Shouldn't it be set to train.mean?Taking a closer look at the measure I saw that the default setting for
aggr = test.mean
is passed.If you agree that this is a bug, than I would fix this. Additionally, I think it would be great if the note does also tell that the time is measured in seconds (right now it does only say: "Time of fitting the model. ")
The text was updated successfully, but these errors were encountered: