#170 Initial commit of two datasets extracted from the Nikuradse paper (Flow in Rough Pipes). - Githubissues

EpistasisLab / pmlb

PMLB: A large, curated repository of benchmark datasets for evaluating supervised machine learning algorithms.

https://epistasislab.github.io/pmlb/

MIT License

805 stars 135 forks source link

#170 Initial commit of two datasets extracted from the Nikuradse paper (Flow in Rough Pipes). #171

Closed gkronber closed 1 year ago

gkronber commented 1 year ago

For issue #170

trangdata commented 1 year ago

Thank you for contributing these datasets, @gkronber. 🚀

Any chance you could merge EpistasisLab/master to this branch again, please? It seems that the dataset update action didn't run on this fork, so I (hopefully) fixed this.

gkronber commented 1 year ago

@trangdata thanks for looking into the failed workflow. I merged master

trangdata commented 1 year ago

@gkronber the update dataset metadata action should have run on your heal-research:170_Nikuradse fork when you added the new datasets. I'm not sure why it did not. Any chance you disabled GH Actions on the fork?

trangdata commented 1 year ago

Thank you for merging again @gkronber. So it looks like this action ran! However, it doesn't detect any changed dataset. I'll have to look back at the code.

Run python -m pmlb.update_dataset_files
changed datasets: []
changed metadata: []

trangdata commented 1 year ago

@gkronber Could you create an empty commit to your branch heal-research:170_Nikuradse and include in your commit message [update_all_datasets] please? So:

git commit --allow-empty -m "[update_all_datasets]"

trangdata commented 1 year ago

This is very peculiar! Not the behavior I was expecting at all. The action shouldn't overwrite the manual reviews of banana and titanic. This is likely due to the reviewers not adding their name to the first line, but still, it should update the metadata.yaml file for your two new datasets. Thanks for being patient @gkronber. We may need to just close this PR and redo.

gkronber commented 1 year ago

@trangdata no worries, I'm ready to continue with this.

Could you please be a bit more specific. Do you mean we close this PR, I branch master again in my fork and then I redo my first two commits to add the dataset?

Please let me know when you are ready with your changes so that the workflow does not change existing datasets.

trangdata commented 1 year ago

Could you please be a bit more specific. Do you mean we close this PR, I branch master again in my fork and then I redo my first two commits to add the dataset?

So I think we would close this PR, merge EpistasisLab/pmlb/master to heal-research/pmlb/master (after my minor fix in #174), checkout a new branch, say, heal-research/pmlb/Nikuradse, add your two datasets. Once you add these datasets, an action should run to create metadata.yaml for each dataset. I will review the PR.

Before we proceed, though, could you help me understand a little more: did you change anything in your repository settings at heal-research/pmlb? I'm still unsure why the update dataset metadata action didn't run the first time.

gkronber commented 1 year ago

Before we proceed, though, could you help me understand a little more: did you change anything in your repository settings at heal-research/pmlb? I'm still unsure why the update dataset metadata action didn't run the first time.

You guessed correctly that github actions were deactivated. I had to manually activate them.

gkronber commented 1 year ago

We retry via a new PR.