Measure Dependency View (similar to the Query Dependency View)
The newly introduced Query Dependency view is phenomenal but what we really need is a Measure Dependency view. Here is what it might look like: http://ppvt.pro/2eXKndh
And a video detailing this idea: https://youtu.be/sYjBB_Ef66M
In one of my production models I have 60 Queries, but I have 200+ measures! I believe that is typical; usually I build Queries and Measures as lego blocks, combining simpler ones into more complex ones needed in the end. In fact I needed it so badle that I attempted to build a Measure Dependency view using NodeXL. While that works, it's a lot of heavy lifting. How cool would it be, to just click a button on Power BI Desktop and see all your measures in a Dependency tree!
This kind of functionality is crucial for building complex reports!
Eugene Niemand commented
More than a year later and this is still the best tool out there https://github.com/otykier/TabularEditor
Anthony Newell commented
Great idea, where is it?!
Impact Analysis, dependency, data lineage - all about tranparency into the data, understanding and trusting the data used in analytics. Enhances users' ability to create robust, meaningful analytics and drive business conversations and decisions.
Gerhard Brueckl commented
Once we have the dependencies, they should also be used when it comes to updating calculated tables/columns after changing e.g. a measure.
Changing a measure should not trigger a recalc of a calculated table where that measure was not used at all // has no dependency to
Brahim Chabane commented
It looks great.
Bryan Plantes commented
An addition to this idea is not only to view the dependency tree for the calculations, but also to view the visualizations dependent upon a measure. This would be extremely helpful for testing if a calculation changes and you need to regression test all the visuals within a workbook which contain those measures. This would also be helpful in getting rid of measures that are not used in your workbook
This could work similar to the "View Dependencies" right click menu selection in SQL Management Studio where you can see the tables, view, triggers, functions, and stored procs that are dependent upon the object or all of the tables, view, triggers, functions, and stored procs your object is dependent on.
And while we're at it could we have a Visuals dependency view as well?
It looks great.
Can you share the source for the recursive function ?
Mimoune Djouallah commented
there is a third party software that support that
Hugo Vanier commented
I did a recursive function for build a dependency tree for each measure. Look at the result:
Xavier Rovira commented
This ia great idea when you develop a model!!! But there is another one, when you explore a model. Create Calculated Fields groups in the fields tab. When the number on calculated metrics increases it is almos impossible to remember the measures you already have. The filter helps but is not enough. I would like to see something were I can group all metrics. Like this
Current Month Signed $
MTD Signed $
QTD Signed $
Variance to Budget (VTB)
Variance to Forecast (VTF)
Sales Forecasting Reliability
Forecast Accuracy %
In my model, just to describe the different status, steps and time calculations linked with one Company "Sales", I have developed more than 250 different calculated metrics. It is tough to see from dependency perspective, and from presentation perspective
Zsakul Iksnibezrt commented
Yes, Yes, and Yes. Necessity when you have a lot of measures in your model.
German Henriquez commented
Now that I have 16 measures I consider this necessary. But I can control it yet.
Onboarding User commented
Had posted same idea some months ago :
Glad to see this thread has more support.
I'll add a link from the "old" idea to this one so we can collect as many vote for this super important feature.
Tom Day commented
Here here!! I have 600+ measures in my staging model and it's very difficult to debug at this point.
Harnidh Singh commented
Great idea Avi. I have been waiting for this big time.