locked
Basic KPI definition. RRS feed

  • Question

  • Hi,

     

    I have a very basic requirement. I have a tabular data source that contains data that represents all transactions that have run through the system. One of the columns in that data source indicates whether or not an error has occured. I can create agregates that represent total transaction count and error count using the micro cube editor. Now, I want to be able to create a VERY simple KPI that reports my "error rate" to the user. Essentially, all I want is to take the (Error Rate / Total Count) and compare the result to a range of values from 0 to 1. Is there a way to do that in PPS? If not, can the KPIMemberValue and/or KPIValue classes be extended so that I can write my "kpi algorithm" to produce these results?

     

    Thanks!

    Wednesday, April 30, 2008 5:50 PM

Answers

  • In v1 you can only enter expressions in MDX for OLAP data sources.  If you can add a column to your data source to do the division there and then just utilize that column for a KPI in your scorecard that would be the easiest and best option you have available today. You may be able to utilize the SDK as well but that would require much more effort. We are looking at options for these capabilities in future releases.

    Thanks,

    Alyson

     

     

    Wednesday, May 7, 2008 2:13 PM

All replies

  • In v1 you can only enter expressions in MDX for OLAP data sources.  If you can add a column to your data source to do the division there and then just utilize that column for a KPI in your scorecard that would be the easiest and best option you have available today. You may be able to utilize the SDK as well but that would require much more effort. We are looking at options for these capabilities in future releases.

    Thanks,

    Alyson

     

     

    Wednesday, May 7, 2008 2:13 PM
  • Adding a column at the data source level isn't really an option for a number of different reasons, major one being the fact that both the numerator and denominator for the operation are aggregate values (count(Error) / count(*)), if this is done at the source, and the data is than filtered based on a dimension, the results would be incorrect, since the filtering would happen in PPS and not the original data source.

     

    I am interested in the ways this can be accomplished at the API level, so if you have a lead of how to do this with the SDK, I'd love to hear it.

     

    Thanks!

     

    Wednesday, May 7, 2008 3:56 PM
  • All of our SDK documentation is available here:

    http://msdn.microsoft.com/en-us/office/bb660518.aspx

     

    Thanks,

    Alyson

     

    Wednesday, May 14, 2008 2:50 PM
  • I'm aware of where the SDK is. You previously stated that "You may be able to utilize the SDK as well but that would require much more effort". I've read through the SDK and didn't see how one would accomplish that, if you know of a way, please advise.

     

    Thanks!

    Wednesday, May 14, 2008 4:09 PM