nci / scores

Metrics for the verification, evaluation and optimisation of forecasts, predictions or models.
https://scores.readthedocs.io/
Apache License 2.0
47 stars 15 forks source link

Is nbviewer still needed? #375

Closed Steph-Chong closed 1 month ago

Steph-Chong commented 3 months ago

There is at least one instance where a tutorial doesn't render properly in nbviewer. (In this case, the first figure in the isotonic regression tutorial, at the end of the first paragraph - see: https://nbviewer.org/urls/scores.readthedocs.io/en/latest/tutorials/Isotonic_Regression_And_Reliability_Diagrams.ipynb).

Now that the tutorials are rendering in readthedocs and binder, in addition to Jupyter Lab, is nbviewer needed as a fourth option?

While I think it is great to have options to meet different needs, if it is possible to remove nbviewer, it would be one less thing to have to check on each merge request. (The tutorials can render differently in each environment).

tennlee commented 3 months ago

I'm not sure whether it's needed or not. The binder page pulls in nbviewer while it's loading, so perhaps it's part and parcel of the binder functionality. I'll need to check on the details for this question. Other than the binder integration, I don't see a need for it given the nice render in readthedocs.

tennlee commented 3 months ago

I've investigated and discussed with others. Nothing would break if we remove nbviewer from our docs, but some people may like using nbviewer.

Steph-Chong commented 3 months ago

If a link to nbviewer is still needed in readthedocs, then no worries.

I found another nbviewer rendering issue. This one is in Flip Flop Index (https://nbviewer.org/urls/scores.readthedocs.io/en/latest/tutorials/Flip_Flop_Index.ipynb) - the first reference has been incorrectly split across 2 dot points

(Note: I haven't thoroughly checked every notebook, compared to a version running in JupyterLab - so I haven't done a comprehensive check. I just happened to notice the rendering issues in nbviewer for "Isotonic Regression and Reliability Diagrams" and "Flip Flop Index").

tennlee commented 1 month ago

Fixed by #615