Numeric data doubles every time TI process is run

[ViRa]

Hi all,

I have a built a cube that loads data from ODBC datasource. The ($) amount values in the measure dimension have been set a Numeric and the date values as String. In addition, I created a view that zeros all the elements for all the dimensions of the cube. TI process has been written to load the cube. In the Maps tab, I chose to 'Accumulate Values' for correct calculation integrity for numeric cells since there are multiple records for the same cell in the datasource and I want the cube to show the sum of those records rather than showing just the latest value (if Store Values option is selected). In the Prolog tab, I used the ViewZeroOut function so the cube is clear before it hits the datasource. To overcome the issue of building up the string values (since Accumulate values has been selected) every time the process is run, I made necessary modifications in the Data tab using CellGetS and CellPutS functions. The cube does not have any rules written since there are no calculations and the cube should simply load the data as per the datasource.

On running the load cube TI process, it completes successfully. All the data appear fine and match as per data source. However, the next time the process is run (without any changes made), the numeric values keep doubling. String data appears fine though. I've tried my best and writing this question on forum because I'm stuck and cannot see any solution. Please guide me as to why the numeric data keeps doubling. Appreciate your time and help.

Thanks

[tomok]

Does it triple if run 3 times, quadruple if run 4 times. etc.? If so then your ViewZeroOut function is not working. Most likely because the view has not been constructed to include the correct elements of each dimension (assuming you aren't just using all elements from each dimension).

[declanr]

Perform a zero out process in the prolog of your ti.
I read your post as suggesting you have a separate zero process and am assuming you don't run it between loads.

[ViRa]

Thanks Declan and Tom for your replies. I'm working on the suggestion by Tom. I did run ViewZeroOut in Prolog Declan. I'm modifying it as per Tom's suggestion. The process is taking a while to load as it is retrieving lot of data and hence I'm not able to update you yet. I will let you know once done.

Thanks for your time.

[EvgenyT]

Thanks Declan and Tom for your replies. I'm working on the suggestion by Tom. I did run ViewZeroOut in Prolog Declan. I'm modifying it as per Tom's suggestion. The process is taking a while to load as it is retrieving lot of data and hence I'm not able to update you yet. I will let you know once done.

Thanks for your time.

Why dont you do ViewZeroOut followed by ProcessBreak in the Prolog? That will give you a visibility whether target data gets cleared out first...
Also may I suggest you try to load a smaller data set for testing purposes? Sometimes you end waiting all this time just to find out your process is wrong

[Wim Gielis]

Also my I suggest your load a smaller data set for testing purposes? Sometimes you end waiting all this time just to find out your process is wrong

Indeed, we all write pyto's from time to time.

[ViRa]

Hi all,

I worked on the suggestion by Tom and ensured I have selected correct elements for each of the dimension in the ViewZeroOut and it worked. That is, the data is no more doubling, tripling and so on whenever TI process is run.

Thanks so much for your help.