topic Community Detection in QlikView

Community Detection

Wed, 25 Nov 2020 16:16:04 GMT

Hey everybody,

Is there a Qlik way how to detect groups within a set of data? Let's say you have 100 people working on different projects. How can I visualize who works together on which task?

Thanks a lot!

Re: Community Detection

swuehl — Thu, 07 Apr 2016 20:20:31 GMT

Not really sure what you are asking for.

In general, the relations should be visible just using Qlik's associative DB in, i.e. if you select a task, you will see all related e.g. people and non-related people will be removed / greyed out.

If you are asking for a special kind of visualization, maybe like this?

Informative, Curvy & Never Before Seen (by me at least) – The Associated Curve Chart | QVDesign

or this

Brilliant Qlikview Parallel Co-Ordinates Chart | QVDesign

(but both are rather complex charts to create)

Re: Community Detection

swuehl — Thu, 07 Apr 2016 20:42:27 GMT

If you are looking for special visualization, here are some more examples using extensions

Visualization of Graph Data

But again, all these and the development aspects behind are not really chart objects for starters.

Re: Community Detection

Thu, 07 Apr 2016 21:01:56 GMT

Thank you!

Re: Community Detection

engishfaque — Fri, 08 Apr 2016 13:52:59 GMT

Dear QlikNewBy,

Please check,

AND and OR

And Mode - YouTube

Kind regards,

Ishfaque Ahmed

Re: Community Detection

swuehl — Fri, 08 Apr 2016 14:48:03 GMT

You are welcome (though still not sure which part of my answer was valuable to you).

If your question is answered, please consider closing this thread by marking helpful or correct answers.

Qlik Community Tip: Marking Replies as Correct or Helpful

Re: Community Detection

MarcoWedel — Sat, 09 Apr 2016 01:48:17 GMT

Hi,

you could present the relation between projects and project members as a pivot table:

In order to accentuate groups of people working in similar projects as well as projects sharing the same members you could sort the project and project member dimensions accordingly:

Please find attached a (rather inefficient 😉 example application.

I tried to create a score that accumulates the differences between neighbouring rows and columns to find a sequence with minimal changes thus creating groups in the chart.

Maybe someone can come up with an optimization algorithm that performs better than the included brute force attempt to create the order sequence ...

hope this helps

regards

Marco

Re: Community Detection

Sat, 09 Apr 2016 07:55:56 GMT

Thank you very much!!

Re: Community Detection

MarcoWedel — Sun, 10 Apr 2016 14:32:41 GMT

You're welcome.

Please close your thread if your question is answered or explain if the proposed solutions do not solve your problem.

Thanks

Regards

Marco

Re: Community Detection

Wed, 13 Apr 2016 20:08:40 GMT

Since I am using Qlik Sense Desktop now, I am unable to open your qvw.

How can I make an "x" in the pivot table?

Re: Community Detection

MarcoWedel — Wed, 13 Apr 2016 20:38:24 GMT

Hi,

hope this helps:

regards

Marco

Re: Community Detection

Wed, 13 Apr 2016 20:45:28 GMT

Hi Marco,

I do not have this template since I am using QlikSense, but if I uae the formula as key figure in my pivot table, it would say that I inserted the wrong fieldname.

Regards

Re: Community Detection

MarcoWedel — Wed, 13 Apr 2016 22:21:32 GMT

try to substitute with your actual fieldname.

regards

Marco

Re: Community Detection

Thu, 14 Apr 2016 06:35:43 GMT

I of course did. 😉

Is there a certain way how the data has to be provided in order to be able to use that formula?

Re: Community Detection

Thu, 14 Apr 2016 19:34:11 GMT

Found it myself: x instead of 'x' (that was easy 🙂 )

How did you sort the table accordingly? With a formula?

Re: Community Detection

MarcoWedel — Thu, 14 Apr 2016 21:58:29 GMT

Hi,

I used sorting expressions like this:

SubStringCount(Left(ProjMembPermut,Index('/'&ProjMembPermut&'/','/'&ProjectMemberID&'/')),'/')

SubStringCount(Left(ProjPermut,Index('/'&ProjPermut&'/','/'&ProjectID&'/')),'/')

Please find attached a Sense example.

hope this helps

regards

Marco