How to exclude unwanted Labels displayed in Dashboard Column for Story?

[Vinodh Chapre]

For Example: In Story there are 3 different Labels=ABC,Labels=XYZ and Labels=MNP. creating JQL Filter Project=Project AND Labels=ABC, In Dashboard it should display only Lables=ABC, but it displays 2 additional Column with Labels=XYZ, Labels=MNP

[Petter Gonçalves · Accepted solution]

Hello Everybody,

Just would like to add a quick update to this thread:

The possibility to select which labels should be displayed in the Two Dimensional filter Gadget (between the issues selected by the filter) is something already requested back in 2013, however, this approach was not implemented due to the low number of requests and the displaying restriction that could cause problems and confusion if not well explained, as described by @Nic Brough -Adaptavist- in the comments above.

You can check the feature request related below:

Add option to the 2 Dimensional Gadget for manipulating labels 

That being said, we understand that we have a significant amount of users asking for this feature today, so we created the following feature requests to properly consider this implementation again:

Select which labels should be displayed in the Two Dimensional filter gadget - Jira Cloud 

Select which labels should be displayed in the Two Dimensional filter gadget - Jira Server 

Feel free to vote and watch the suggestion to increase its priority and also receive notifications about any updates.

[J Khan]

Awesome, thank you Petter! Hope the feature gains traction and users see the value. 

-Regards

[J Khan]

If this feature appeals to you, go vote for it: note both server and cloud proposals are out there -->

https://jira.atlassian.com/browse/JRASERVER-71790

https://jira.atlassian.com/browse/JRACLOUD-75527

[jfuentes]

Hi But in reality this not help exclude labels only create a white list of Labels.  So what if the task have various labels and you only want to exclude one.  Are you saying I have to write a syntax for every label I want included?  That would be annoying if I have lets say 30 labels and only wanted to exclude 1 then I would have to write 29 AND Label syntax parameters. 

[J Khan]

Excellent jfuentes! Agree!

[J Khan]

If this feature appeals to you, go vote for it: note both server and cloud proposal are out there -->

https://jira.atlassian.com/browse/JRASERVER-71790

https://jira.atlassian.com/browse/JRACLOUD-75527

[Vinodh Simon]

Hello Team,

    Jira tickets are used by multiple teams, and each team wants their own label and want to view the graph based on the labels used by their teams.

    This is limiting the use of Jira Tickets to a single team. Please update the Gadget, where I should be able to choose my labels.

[Nic Brough -Adaptavist-]

Please see responses above.

[Nic Brough -Adaptavist-]

I'm not saying anything that's wrong, just telling you why the request may be counter-intuitive as most people don't expect to ask to see a list of issues and have parts of the data arbitrarily redacted for them.  I've just stated that's why Atlassian will probably never build such a gadget.

Your solution would work better than the others suggested here, as it wouldn't be as confusing as a report on a load of issues that you've asked for that then doesn't display the issues properly.

[Joe Schaefer]

Yes, I am having a variation of this same issue as well. After failed attempts, I had to google if it was possible, and I am very glad I found this thread.

More than 4 years from the start of this thread and we still do not have this functionality! I hope this is something that could be looked into in the future.  

[Nic Brough -Adaptavist-]

It displays the labels there because they exist and you've configured the dashboard gadget to show you labels.

I'm not sure what you're trying to do here, but could you start with explaining "dashboard column" (what gadget?  or what report? etc) and then what you'd like to see from it.

[Vinodh Chapre]

JQL: project in ("EMI - Catalog", "EMI - Product and Item", "EMI - Load/Sync", "EMI - Brand Management") AND labels=Team_Hermes AND Sprint in (openSprints()) AND issuetype in (Story, "Support Ticket") AND cf[10211] != Spike

Gadget: Two Dimensional Filter Statistics
XAxis: Status
YAxis: Labels

Above JQL does not pull only Team_Hermes, if there are more than one Label for that Story, it pulls all the Label and shows in Dashboard as extra Rows.

[Nic Brough -Adaptavist-]

Ok, yes, that's correct.  It's showing you the labels on the issues you have selected.

[Vinodh Chapre]

Hi Nic,

Can you please let me know how we can avoid other Labels.

[Nic Brough -Adaptavist-]

You don't.  They're data on the issues you are reporting on.  So when you display the labels field, you get to see the values in the field.

[Thierry Dalon]

@Nic Brough -Adaptavist-

I also would like some gadgets like this.

For example I want to display the issues with a label matching "Loc_*" and in the gadget (either pie-chart or table) only display these matching labels.

(Here the use-case would be I want some location based reporting with labels like Loc_Paris, Loc_Berlin ....)

Anyway to achieve this? Maybe have to go to a custom Confluence Report?

Many thanks

[Nic Brough -Adaptavist-]

It's the same answer - your report says "show me the labels", so it shows the labels on the issues that match "label = loc_*"

[Thierry Dalon]

I know the current gadget are working "as specified". But still you are not answering to my need.: how to do get what I am requesting. (Same answer does not help)

[Nick Sanders]

I have the same question. All the issues to be displayed have two labels, The first indicates the group owning the issue eg Engineering, QA etc and one indicating the functional area where the issue has been reported eg Video, Audio etc. I want to show the functional area only on a pie chart and hide the group information.

[itikka]

We have the same question. For our project, we use different types of labels. We would like to display subsets of these labels (either specified by the names of the labels or by regex) in a pie chart on the dashboard. How can we do that? If it is currently impossible, can this functionality please be added? (The above replies do not answer this question.)

[Kaycee Neoh]

I face the same question too. It looks messy when I try to create dashboard and it actually list out all the labels including the ones I don't wish to see (although it's labeled to the issue)

Hope to have this function and it would help a lot!

[Sandra O'Leary]

Same here, I have the same need for this type of gadget.  I have labels for customer type and labels for function.  If I want to see the issues by customer type and I use the 2 Dimensional Filter Statistics, I have to scroll past all the other labels that I don't want to see in this context.

[Michael Harrison]

What I don't understand is how pie charts keyed to the label field decide which label to display for an issue with multiple labels.

For instance, I have 2 issues. One has a single label "A" and the other has "B" and "C".

Will the pie chart be split in half, with two issues illustrated? One side is "A" and the other side is... "B" or "C"?

Or will the pie chart be split into thirds, with a slice for "A", "B", and "C"? What will the Total Issue count be?

[Nic Brough -Adaptavist-]

It's a third - you've got three labels to display counts for.

[Naomi Burkhart]

This link helped me https://community.atlassian.com/t5/Confluence-questions/How-can-I-show-statistics-on-labels-used-in-Confluence/qaq-p/402889

You have to create a macro using that code, then use that macro in the entries of a table, and use the macro "create chart from table"

[Luis Garrido]

Hi there! Any update on how to exclude labels in a 2-dimensional Jira gadet? Any script or query that I could use to manipulate the output of my filters in Jira? Thanks!

[Radhika Darbari]

Hi, This is also a problem for me. Sames as people above I want to find a effective way to just see certain label stats, not all 'Labels', can Atlassian help make a filter for just 'Label =xxx'  and then if that filter created hide others labels attached to those issues in a widget? 

[Ananth K]

I have same issue. JQL filters labels you want , however the issues may have other labels too.  I want to see all these issues on my dashboard but want to display grouped by only labels of my interest and not other labels.  This is definitely very useful use case for Atalssian to address.

However, if community here knows way to do it , it would be great ! 

[Nic Brough -Adaptavist-]

There is no way to do it directly - the gadget is given a list of issues, and asked to display the labels on those issues. 

If you don't want to see certain labels in there, then you have to remove them, either from the result set, by excluding them completely (and hence losing issues that have other labels), or remove the unwanted labels from the issues.

The only way to do what has been described is, as before, write your own gadget that knows which labels it should not display (ideally with a config option for it, rather than hard-coding)

[Jannis Kantsaporidis]

Thanks for your response Nic. What is required in order to write my own gadget? I'm not sure what type of Atlassian services the company that i work for has, besides using JIRA that is.

[Nic Brough -Adaptavist-]

Some knowledge of java, and usually some html.  But I would not bother doing it from scratch.  Atlassian have Software Development Kit that provides a load of useful functions (At the most basic level, if I want to write a new Atlassian app for Server, I run "atlas-create-<jira/confluence/etc>-plugin", answer four questions, then running "atlas-run" builds the app, fetches a copy of the system I'm building for, installs the app and runs it, so I can go straight into testing what I've written)

There's loads of developer docs too, but in this case, I would start at https://developer.atlassian.com/server/jira/platform/writing-gadgets-for-jira/

[jtisge]

Hi @Vinodh Chapre  I used below filter to avoid diplicates with respect to labels.

 "   labels in (A,B,C)) AND (labels not in (D,E) OR labels is EMPTY) " .   This way I avoid tickets with labels D and  E.

[Karthika Menon]

this helps.Thankyou

[rajanibs]

How is that? this query will remove records with labels D and E or not having any labels, basically the query will give only records having labels A, B, C. Issue raised was about dashboard display of records having multiple labels, but wanted to show only a few required labels.

[Nic Brough -Adaptavist-]

Yes, it's broadly wrong, it does not "avoid duplicates", it drops a load of stuff the people probably do actually want.  

There's two failures in this search:

• the "or labels is empty" is not much of a failure really. The "labels in (A,B,C) and" already means labels cannot be empty.  It won't break anything, it's just a waste of space and potentially confusing.
• the important failure "and labels not in (D, E)" means that issues that could have labels of A, B and/or C will be dropped from the results.  Imagine issue XYZ-123 that has labels B and D.  It will not be reported on by this query, despite having the label B that is what the user is looking for.

[Ela Gavrilova]

Adding my +1 to a request for this functionality! Since the filter is built to target specific labels, I'd love to also have the opportunity to only display those labels on my graph/pie (:

[Nic Brough -Adaptavist-]

Do not mix up "filter results" with "what I want to display".  If you tried to do this based on a filter, it will fail for your humans.

Imagine a simple case.  We issues logging what pets people live with, and we have three current labels:  cat, dog and other-pet.  Now imagine we have several households:

• Nic lives with two cats
• Carol lives with two dogs
• Nes lives with a dog and a budgie
• Andrew lives with a dog and a cat

A search for "label = cat" should always return Nic and Andrew.  "label = dog" must return Carol, Nes and Andrew.  "Label = other-pet" must return "Nes".  If the filter does not return those, it is useless because the answers would be wrong.

You must look at the reporting.  You'll need to find or write some form of "exclusive reporting" - it will need configuration that says "ignore loads of data and only report on one thing"

This is quite limiting in real life - I can't think of a decent use case where "Carol, Andrew and Nes live with a dog" is more useful than "Cats live with Nic and Andrew. Dogs live with Carol, Nes and Andrew.  Nes has another type of pet"

[Ela Gavrilova]

In our case, labels are used for multiple different purposes. The report I'm wanting to generate is based on a subset of labels only, to measure the frequency percentage for each label. This is undermined by the presence of other labels in our dataset, although those labels are integral for other processes. The result of the report is skewed, because it shows a completely unrelated label present on 100% of tickets and thus being very common - it masks the frequency of the labels I'm actually trying to report on.

[Nic Brough -Adaptavist-]

I can't see how this is "masking" anything.  The filter for "label = x" returns the issues that have label x.

The filter reutrns a list of issue that have x.  You seem to be focussing on "issues that have x, but also y".  The filter is nonsense if it does not report all issues with X, so, ff you're looking at other labels and conflating them with x, then just stop looking at them.

[Ela Gavrilova]

To clarify - the filter returns all issues that I'm looking for based on the labels I'm looking to track. Let's say those are labels xx, xy, and xz. These tickets have additional, unrelated labels on them - let's call them aa and ab. All tickets have the aa label, and half of tickets have the ab label, and then xx, xy, and xz, are approximately evenly distributed throughout, with some duplication. 

I'm looking to create a pie chart that shows which label out of xx, xy, and xz is most common. I ask for a pie chart based on this filter that shows me the incidence of label by volume. It will tell me that the bulk of my labels is aa, and then second most common is ab, then showing xx, xy, and xz as not very common. This is not an accurate representation of the prevalence of the x labels. As well, the audience reviewing these pie charts has no connection to, understanding of, or use for the a labels, but they're shown that they're super common among the filter results. "Stop looking at them" isn't helpful when they're looking at a percentage of irrelevant. 

What I want is a filter that shows the user only the x labels and their prevalence on the tickets in the filter. 

[Nic Brough -Adaptavist-]

How is it "not accurate" to report on exactly the data you have?

You need to stop mixing up the search with the report.  You have searched for (and saved as a filter) "everything that has label X".  You have then chosen to report on that with a gadget that adds up the number of instances of each label.

There is nothing wrong with your search, the problem is that the reporting is not what you want. 

What you really want to do is stop reporting on stuff you don't want in there.  This means finding a report that enables you to stop looking at things.  "Stop looking at them" might not seem "helpful" here, but it is the only accurate answer.

Also, a pie chart is utterly useless as a report for the question you have.  If you want to report on the "prevalance of a label", a chart will just give you 100%.  What you really want is a chart that compares the prevalence of labels.

Jira has that.  What it does not have is "and stop looking at some of the data", because there's very little call for it.  Most of us are perfectly happy with pie chart that shows us an accurate representation of everything, we can ignore the segments we don't have an interest in.

[Ela Gavrilova]

>"very little call for it"

I may be misunderstanding this forum topic, but is the request in the topic not a call for doing exactly what I'm asking for here?

> "Also, a pie chart is utterly useless as a report for the question you have."

It isn't entirely useless, from what I've found. It's challenging to describe what I'm looking at without a visual. The pie chart functionality gives me most of what I'm looking for in that it will provide a visualization of the below data set, (returned by a filter that looks for any tickets that have labels in xx, xy, xz):

ticket #1, labels: xz, xx

ticket #2, labels: xz

ticket #3, labels: xy

that says "on the tickets within your filter result, 25% of the labels are xx, 25% are xy, and 50% are xz" - exactly what I'm trying to show.

If, however, the label "aa" is also present on all of the tickets with the same query as described above, as shown in the below data set:

ticket #1, labels: xz, xx, aa

ticket #2, labels: xz, aa

ticket #3, labels: xy, aa

I then see that there are 7 labels total, ~43% are aa (not what I want to see), ~29% are xz, and ~14% each are xx and xy. 

The above is not wrong, but it shows data that I'm not interested in. In addition to this, it also indicates that 29% of the labels are xz, which is significantly less than 50%. A decrease in percentage applies for the other "x" labels. Now, let's say that "x" labels are used to track types of errors and "a" labels are used for triage. A user wants to see how often a particular error is made in comparison to other errors. The true result should show that 50% of all errors made are xz, but instead it appears to show that only 29% of all errors made are xz. This assumption is, of course, inaccurate, given the fact that there is irrelevant information in our dataset, but to a user who glances at the pie chart and expects to see, by percentage, which errors are made and how often (perhaps to determine when coaching needs to take place on a given error), they then assume that the error isn't made nearly as often as it actually is.

From my understanding of this forum topic, it looks like I'm not the only person whose users want this type of functionality, if not specifically for pie charts. 

[Nic Brough -Adaptavist-]

I understand what you're trying to do, but it seems like the report would arbitrarily drop data out, which may not be useful for most.

You could explore writing an app that provides the report you want.  And I would strongly recommend some text in it that tells people that it is ignoring data deliberately, so they don't misunderstand what it is saying.

[MK]

As long as labels cannot be searched by prefix (e.g. "labels = xyz_*"), a filter will be utterly useless. Listing all labels manually is totally inefficient. And why not give users the possibility to display each in itself complete set of prefixed labels directly in a dashboard? But then, likely moving away from JIRA anyway without server version support ...

[J Khan]

Nic, once again, it is not "arbitrary dropping" but selective isolation of data and your comment is evidence that you still do not understand the concept of winnowing important data labels down with a prescribed subset of relationships; I am not interested in 100% of the data, I am interested in a snap shot of a subset of data aligning to a subset of labels for action in the work process. Not all labels are the same. Good luck to you my friend.

[Nic Brough -Adaptavist-]

I'll give an example:

I put a label on ABC-123.  Later, I go to a dashboard that displays labels.  You, as the author of the board, has decided they are not interested in the label.

They have arbitrarily removed data I expect to see and should be there.  You have arbitrarily dropped something I want or expect. 

This is less than ideal as I would need to check, and could cause me to think reporting is off because it is missing stuff.  I'd suggest being very careful about how you do it.  And why.

[J Khan]

Yes Nic, gosh you're right. My nearly 10 years of using JIRA in large development environments has led me to wrong conclusion on purpose built dashboards. Thank you for showing me the error in my ways. I do however appreciate the warning about viewer expectation of data, thus the request for a 'none of the above' category as suggested by another user in the Issue request. Can we stop this now please and agree to disagree, I don't have time to continue wasting.....

[Nic Brough -Adaptavist-]

From someone with 16 years of using, deploying and supporting Jira in large development environments, and seeing this sort of thing causing all sorts of problems, I'm speaking from experience.

Seriously though, the main problem I've seen is when people try to do this sort of thing, they fail to consider others around them.  It's fine when you do it, because you understand it, but others will not.

[J Khan]

Not angling for last word but the filter result is already less than 100% of the population right? What's the difference?

[Nic Brough -Adaptavist-]

Correct - a filter result is less than the entire population of issues (unless you run a blank query, then it's "all the issues you have access to, which may be all of them").

I'm not sure what you mean by "difference" though - a filter result is just the issues, it's not the data used to find them, or the data on them.  The difference between that and what?


Just to be clear, it's worth explaining something that was misunderstood before in this conversation.

Generally, in Jira speak, some of words are mixed up and often don't mean the same thing to people and that gets in the way of conversation sometimes.  So, I'm going to say:

• Filter: This is saved question, done in JQL.
• Filter results:  A list of issues you get when you ask the question.  Note - this is just a pile of issues, nothing more.  It has no output, no result, no content in itself, it's just a list of issues
• Report:  Something that reads a list of issues and fetches the data from them that it is coded to need so it can (usually manipulate) and display it

The "ask" here is to arbitrarily supress the display of some of the data that is on issues.  Reports that report on raw field data all show you all the data in that field.  So if you ask to see a labels field, you see all values set on the issue.

Much of the conversation was originally about the problem that people were demanding to do it in the filter, which is physically nonsense, because there's no data in the filter.  The filter is "give me a list of issues that match this rule", not "and drop some of the data off those issues".

It would have to be done in the reporting layer, during its manipulation of the data it takes from the the issues, as that's the layer that displays (or not) the data.

My problem with clarity here is that when I ask something to show me a field, I would expect it to show me all the content of that field.

[rajanibs]

X or Y axis on the two dimensional dashboard to show the values that field has and you cannot control on what should be shown there, there is no provision to achieve what you are looking for. This is applicable not only to the labels, but to any other fields. If the field has multiple values and you want to show that in either X or Y axis, it will display all available values.

It is not a problem to solve, at least till now :)

[Steve Dix]

Is there a gadget that allows me to apply a filter to the results that show in the dashboard.   I too would like to hide certain labels.   I understand that If I have 5 issues with lablel "A" and 5 issue that include both label "A" and "C", that today I will get a dashboard with 10 "A" results and 5 "C" results.  What I want to do is hide those "C" results on my dashboard counts.   Similar to hiding in many spreadsheet table features.

Is there a gadget that allows for this?

[Nic Brough -Adaptavist-]

I have not found one yet.

[Jannis Kantsaporidis]

Does anyone know any work around for this problem? Again as previous posters have stated, I want to exclude specific labels from the issue field so that they aren't presented in the Y or X axis on the two dimensional dashboard.

[Sanderson]

I have the same issue. I have a number of labels I want to show and a number of labels I don't want to show. Are there any available gadgets right now that are able to select for labels? I have spent quite some time researching this but I'm really not finding something I can use.

[Nic Brough -Adaptavist-]

There's still nothing new, since 27th September comment on the previous answer