Is it possible that the interactive bar chart node has a bug?
I’m plotting the returned quantities from several customer (category) filtered by year.
Everythings is ok if I select a view where all years are selected, if I select the data visualization on the interactive bar chart by single year, there are no corrispondence between the category on the x-axis and the real Bar data: i.e. on the x-axis in corrispondence of the customer “A” there is the bar corresponding to returned quantitie in the selected year of customer “D”.
Is it a bug or a my error?
I’m having a hard time understanding the problem based on your description. Could you upload a small workflow and dataset that shows what you’ve tried so far, and add some screenshots highlighting what you think is wrong?
Hello Scott,
I understand that it is difficult to understand especially with my very bad English! I’m sorry but I cannot share dataset and relative workflow as there are confidential data and informations of my company, by the way I built a simplified picture that describe the problem, see below.
I experience this problem with all the bar charts I generated.
The view is correct for the year 2019, but moving to 2020, 2021 tec. the views are not correct.
So if I understand you correctly, when you deselect categories for your bar chart in the legend that correspond to calendar years, the mouseover value shown for the classes is incorrect.
I’m not able to reproduce this on my side, so I can’t tell yet if this is a bug or something specific to your own configuration. For example, as best I can tell, this is all correct:
I individuate the problem, it stays in the data set!!!
The data set generated trough data cleaning and pivoting is constituted by row where there are some missing value. In that case deselect categories for the bar chart in the legend that correspond to calendar years, the mouseover value shown for the classes is incorrect.
If I substitute the missing value with “zero” the bar chart work well! I don’t fully understand why but it seems to be the problem.
Sounds like missing values are not being handled properly then? Interesting.
Can you either 1) repost the sample WF in an executed state or 2) post the Excel spreadsheet? Right now I can’t actually execute the nodes in your WF since I don’t have the file.
Here below you’ll find also the two excel files used for testing bar chart: “Prova” contain the missing values and “Prova1” are the same where missing values has been substitute with “zero”.
Actually I’ll tested the bar-chart node both in “group” and “stacked” configuration and the above reported issue happen only in the group configuration.
Thanks for your patience and for uploading the WF and files. I see the issue now with the labeling of the class when missing values are present (for example, A34 is not A10!)
This definitely seems like a bug and I will bring it to the attention of the dev team. In the meantime, it seems that you have already found a workaround by setting your missing values to zero, which is a good idea in this situation anyway
Not yet - for the moment I was just updating the thread with the ticket number. It will be up to the development team to determine when/if this gets implemented in a future release. If there’s an update we’ll share it here.
I would like to point out that the bug has not yet been solved with version 5.2.5. You should always set missing values to zero to work around this problem.
is there a reason you are still using the Bar Chart (JavaScript) instead of the new views?
Does the bug also exist in the new views? With the new views being out of labs (since 5.1 already), we will likely only fix crucial bugs for the JavaScript views. If I understand this issue correctly you can avoid the issue by replacing missing values with 0?
I’m sorry but I cannot share dataset and relative workflow as there are confidential data and informations of my company, by the way I built a simplified picture that describe the problem, see below.
