Box-Whisker-Plots JavaScript probably has Bug

Dear KNIMErs

I using the Node JavaScript-Conditional BoxPlot for Box-Whisker-Plots which I really find great. If the node is reseted and reinitialised, then the Whiskers are gone and only the Box and the Outliers remain! The Datasets and Node Settings are the same! In the Fileattachements you will see the screenshots of the same Dataset:

 - are pasted are the right one with whiskers

 - and the new one without whiskers respectively.

Any Advice for me? I am grateful for Feedback.

Greatings Sami

Hi Sami,

thanks for reporting this. Unfortunately I cannot reproduce this issue. Can you tell me on which operating system you are? And can you maybe even share a simple workflow were this problem occurs? What do you mean by "reinitialised"? You click the reset button and then execute the node again? Also: Does this problem occur also in the image that the node outputs or only in the interactive view?

Kind regards,

Alexander

Hi Alexander,

I work on "Windows 7".

Exactly by "Reinitialised" I mean the node is reseted und reexecuted, for example if new data acquired from Database. The Problem occurs at the interactive view. It did work before. When I have an old workflow which I have saved 3-4 weeks ago, then the results are displayed correctly. If I reinitialise the same workflow again, or I make new workflows, then I get it without the  whiskers.

I am attaching a workflow as example, the Example Dataset should be inside the "table creator". for any case I uploaded also the Excelsheet.

I appreciate your feedback.

Best regards

Sami

Hi Sami,

I could reproduce the issue and think this is indeed a bug. I will see how this can be fixed and hopefully this will make it into the summer release of KNIME 3.4. Thank you for reporting this!

Regards,

Alexander

Hi Alexander,

also thanks for your Feedback.

Regards Sami

Hi @AlexanderFillbrunn and @sami_elsabagh

Do you know if this bug was ever fixed or a work around developed?
Also the bug here - the JavaScript-Conditional BoxPlot error no Whisker displayed
I’m still getting the same problem with the images coming from the Conditional Box Plot (JavaScript) but not the interactive ones.
I’m using 3.7.0.

Thanks

John

Hi John,
actually according to my info yes and no! For conditional box plot (JavaScript) it works in both interactive and file Export mode.

For simple box plot still the Whiskers are missing.

Also to be honest I myself have downgraded from 3.7.0 to 3.6.2 due to other Problems, and will stick to it for a while. So I cannot give Feedback for 3.7.0

Hope this helps you further
Regards Sami

image

2 Likes

Hi Sami

Thanks for the speedy response and help.
Unfortunately I’m still getting the problem in the standard boxplot image

But not in the interactive view where the extreme outliers are visible

Sadly it’s the standard boxplot images I need for future steps :frowning:
I will try this in 3.6.2 on a different PC to see if it resolves the bug, I agree 3.7.0 has it’s problems, hopefully all will be resolved soon.

Regards

John

Update

The test with 3.6.2 still failed to give extreme outlier Xs in the box plot image.

John

Hi John, really strange. I think something is really going wrong. I have at both (conditional and Standard JS Boxplots) all outliers both as Image and interactive. Only by the nonconditional boxplot the Whiskers are not displayed.
Just to mention, I got Problems by upgrading from one Version to the other, so I uninstalled and reinstalled everything new.

If it all doesn’t try either the Standard one of KNIME or the snippet R Plot.
Sami
image

1 Like

Hi,
thank you for bringing this issue to our attention. I could not reproduce the issue with missing extreme outliers, but I fixed the problem with the missing whiskers. It should be correct in the next bugfix release KNIME 3.7.1.
Kind regards
Alexander

1 Like

Hi Alexander,
thanks for your Feedback.
Kind regards
Sami

1 Like

Hi Alexander

Thanks for looking into this. Weirdly the whiskers seem fine on my plots it’s just the extreme outliers that are consistently missing in the standard image (but not the interactive).
The Conditional boxplot node (non-JS) works fine but there’s no way to pipe the image further down the workflow, and likewise for the interactive version in the JS node. unless you know of a way.
So look like I will have to use the R plot node Sami kindly suggested.

Is there a timeline for the 3.7.1 release?

Kind regards
John

Hi John,
If I’m not mistaken the new version will be released next week. Would you mind sharing your workflow or at least the part that is creating the boxplot so I can have a look?
Kind regards
Alexander

1 Like

Hi Alexander

That would be great if the new version comes out next week.

Here is the boxplot part of my workflow.example_buggy_boxplot.knwf (12.7 KB)

Let me know if you have any questions or need some example data.

Thanks for looking into this.

John

Hi John,
if it is ok to share, could you also send me the corresponding data? Without it I cannot execute the workflow.
Alexander

Hi Alexander

Here is the same workflow but with example data embedded. It consistently reproduces the bug.
example_buggy_boxplot data.knwf (21.6 KB)

Thanks

John

Hi John,
that’s strange. The outliers turn up just fine in my image (see attached screenshot). Which browser is your KNIME using for generating the images? You can check that under Preferences -> KNIME -> JavaScript Views -> “Please choose the headless browser to use for generating images from JavaScript views”.
I am using the bundled headless Chromium, which is the preferred method.
Kind regards
Alexander

1 Like

Fantastic! Thanks Alexander

That seems to have fixed it. I set to Bundled Chromium from Chrome in the top set of radio buttons and to Bundled Chromium from the Bundled PhantomJS in the 2nd set below (see image below) and it worked.

I’m a little surprised that the settings didn’t default to Bundled Chromium away from the support discontinued PhantomJS with the recent 3.7.0 update; is this something that can be set / driven in future KNIME updates?

Many Thanks again @AlexanderFillbrunn and also to Sami @sami_elsabagh

John