A couple of questions/suggestions for future functionality

A couple of questions for the R&D team:

 

  1. Do you think it would be possible at some point to decouple the Prediction functionality from the Tabular back end?  This way, we could perform predictive analytics with plain old MDX/Multidimensional as the input.
  2. Would it be possible with the Data Discovery designer to enable another axis called Color, that we could optionally drag hierarchies onto?  Given that MDX supports result sets with 128 axes, this seems like it would be an easy method to add some more precision to visualizations, and it would work similar to other tools such as ggplot2.
Reply
9replies
  • 1. Are you referring to the mdx prediction:

    https://msdn.microsoft.com/en-us/library/ms145621.aspx

    Reply Like
  • Hi,

    About question 2, we have logged this as a feature request and will update as it becomes available.

    Tnx,

    m

    Reply Like
    • Tim Andrews
    • Business Intelligence Manager
    • 4 mths ago
    • Reported - view

    Bar Amit 

    Hi Bar.  I was referring to the R integration, prediction algorithms that you have integrated into BI office for Mash Pit modeling.

    Thanks

    Tim

    Reply Like
    • Tim Andrews
    • Business Intelligence Manager
    • 4 mths ago
    • Reported - view

    Michael Raam 

    Great, thanks Michael!

    Reply Like
  • Hi Tim,

    What are you suggesting ? or how would you like to use the existing R based predictions ?

    As each langue has its Niche, R is better suited for data crunching and there is a large user base for R  especially for prediction type analysis.  While MDX is better suited for data retrieval and hierarchical data navigation , grouping and calculations.

    Tnx,

    m

    Reply Like
    • Tim Andrews
    • Business Intelligence Manager
    • 4 mths ago
    • Reported - view

    Michael Raam 

    Hi Michael,

    For example, if I have some new customers in my SSAS DB that have not generated any activity, I may want to predict their potential lifetime value based on a model that I have trained against existing customers.  The only way to integrate this into my existing BI Office Storyboard would be use a separate tool and embed a web asset within Storyboard.  It would be great if I could pass the results of an MDX query into the R engine via your Advanced Algorithms tab.

    Reply Like
  • Tim Andrews

    Hey Tim,

    If your old prediction is saved, you can rerun it via the data discovery.

    Right click on the graph -> Content Options -> Analyze further.

    In the data discovery: Analytics ->Prediction-> menage prediction model, and rerun your old prediction.

     

    Hope that helps,

    Bar 

    Reply Like
    • Tim Andrews
    • Business Intelligence Manager
    • 4 mths ago
    • Reported - view

    Bar Amit 

    Hi Bar.  According to the help docs and my personal experience, in order for the Prediction button to show up, the backend has to be a tabular model created through the Mash Pit.  It doesn't show up at all if using a multi-dimensional backend.

    Reply Like
  • Tim Andrews 

    Hey Tim,

    I miss understood you.

    Pyramid prediction is currently working on Tabular model , I'll add it as a feature request.

    Bar

    Reply Like
reply to topic
Like Follow
  • 4 mths agoLast active
  • 9Replies
  • 410Views
  • 3 Following