Forum Discussion

Tasos's avatar
Tasos
Helper II
7 years ago

Dynamic summarize

Hello,

 

A question that can either be really easy to answer or too difficult.

 

I have a database with Orders; Order numbers, product code, date, quantity sold and various other flags (country, product category, e.t.c ).  Obviously, one order can have one or multiple order lines (based on the different number of products sold).

 

I have created a summary table based on the individual order number. In this table, I have the number of lines and the total quantity sold per order.  In addition, I have created an output based on the different number of lines counting the number of orders (1 Line Orders are X, 2 Line Orders are Y, e.t.c.) 

 

As I understand that it can be difficult to be understood, I have shared an example of the database

https://www.dropbox.com/s/2715sk921l0iw3r/dynamic%20summarize.pbix?dl=0

 

Thanks for your time,

Tasos

 

My problem starts when I apply filters. Let's assume the order 123 has two lines (ProductA and ProductB).  Additionally, we have a slicer for the different product categories. When the end user selects the product category 1, product B is excluded. However, the graph, that in X-Axis uses values from the summary table above, still shoes that the order 123 has two lines.

10 Replies

  • Hi Tasos

     

    Do you think you really need to create this summary table? I do not think you need it, unless there is something I am missing.

     

  • Tasos based on your description, you cannot pass on slicer information to summarize table, either you have to have category dimension in the model and set the relationship with calculated summarize table and use that dimension for slicers.

  • Chihiro's avatar
    Chihiro
    Solution Sage

    What you need is not summarize, but dimension tables linked to fact table to calculate measures based on different segmentation.

     

    Ex: You'd add dimension tables for ordertype, country, calendar etc.

     

    And then link dimensions to fact table (Orders), using column(s) from dimension table as primary and columns(s) from fact table as foreign. This is typically referred to as Star Schema and one of most common ways to structure data model for reporting/analysis.

     

    You can find articles in links:

    https://docs.microsoft.com/en-us/power-bi/desktop-create-and-manage-relationships

    https://businessintelligist.com/2014/11/21/tutorial-how-to-create-a-star-schema-model-in-power-bi-and-power-pivot-using-power-query/

    • parry2k's avatar
      parry2k
      Super User

      Tasos i looked at your pbix, not sure why there is a need of summry table. Can you clairfy ?

      • Tasos's avatar
        Tasos
        Helper II

        parry2kChihiroLivioLanzo

         

        Thank you all for the immediate reply. 

         

        I am not really sure if I do need the summary table or not. The reason why I thought I need it, as I mentioned in my first message,  is that the same order can have multiple products/countries etc. What I want to create is a dynamic bar chart where in the X-Axis I am having the different numbers of order lines and the number of orders is used as a value. (in the pbi file, I have added the wanted graph)

         

        I am looking at Chihiro reply now. Hopefully, my text above helps you understand my original message.

         

        Thanks again for your time

    • Tasos's avatar
      Tasos
      Helper II

      Chihiro parry2kLivioLanzo

       

      I have been trying to understand how Chihiro's suggestion could answer my question and I fail to do that. 

       

      Could you please use the example and make a high-level example of what you meant?

       

      Thanks,

      Tasos

      • Chihiro's avatar
        Chihiro
        Solution Sage

        Oh, I get it now, I misunderstood your question orignally.

         

        To clarify, what you want is dynamic histogram, based on slicer selection?

         

        This isn't possible through conventional means that I can think of.

        Since DAX tables and calculated columns are calculated when the model is first processed, and will not update with slicer selection.

         

        Are you ok with using R-Visual with script?

         

        There may be way to do it in DAX, but I'll have to think on it a bit.