transformations of complex metrics

[tootscamus]

I have created a metric which is quite complex, it relies on 10 other base metrics. I need to produce several transformations of this metric but the transformations do not seem to work. I have been told that I need to create transformations for all of the base metrics in order to make the overall transformations work. Can anyone tell me if this is indeed the case or is there a simpler solution.

Any help is appreciated

 

[nlim]

what type of transformation do you need to use here?

Have you considered putting the base metric formulas into the final formula as one big metric?

 

[tootscamus]

Thanks for the response.

The report accepts a prompt for a chosen month, then displays that month's calculation along with the previous eleven months and a percent change between the last two months. The dimensionality constraint of Microstrategy does not allow me to do this easily. (I could whiz-bang this one out in Business Objects).

I would have liked to have built one big metric, but it just does not seem possible. I am attempting to emulate in Microstrategy a calculation which was originally programmed in Focus. The calculation draws on facts from many tables and requires some metric values to be created from only attributes (target values for determining efficiency). This efficiency testing can only occur after the other calculations have been performed.

Is my assumption correct that, if my metric is built on other base metrics then transformations must be built for the base metrics?

Thanks again,
Toots

 

[nlim]

MSTR i think does require base metrics to be transformed. I guess the good news is that you can manage the objects centrally after you are done. No more changing a column name without knowing which reports or local variables will break.

yeah, business objects has a great excel-like mentality to it. Can almost do everything with it, just hard to manage after some time.

offtopic, why did you folks stop using BOBJ? Were there any technical reasons?

 

[tootscamus]

Thanks nlim,

I agree with you that BOBJ becomes difficult after a while, so much imbedded calculations that are difficult to correct (especially if you were not the one to have created the original report).

I am working as a contractor for a company that has never had BOBJ. I have used it in the past.

I must tell you that I have never posted to this forum before but I have found many answers here to commonly asked questions and you are usually the one with the best responses. For all of us out here, I thank you.

 

[entaroadun]

There are many ways to skin a cat. I personally don't like transformations, and find ways around them.

I'm assuming you have a Month attribute for your time dimension. I would build an R12 Month attribute as a many-many parent of Month. Prompt your users on the R12 Month attribute and return the 12 Months in the report. That way you only need one metric with no transformations.

The only complicating factor is the metric that shows the percent change between the last two months. Does this need to be on this particular report? Are you using Desktop as your end-user interface? If not, and you are using Web or a custom engine, you can write another report and merge it with this one to make it look like one report, even though the dimensionality for the two would be different.