I am stuck... for quite a while

I have a calculated member: forecast_variance in the measures_dim. It is Forecast of (thisYearthismonth - lastyearthismonth). I used the ParallelPeriod function. It works ok

However, when I calculate the % of forecast_variance , I am having a problem. Here is the scenario. In my Prod_dim, I have ProdA and ProdB. Under ProdA, I have SubA; Under SubA, I have A1, A2..

03/2004 03/2003 fc_var % fc_va
ProdA 0 0 0 (-75%) --> wron
SubA 0 0 0 (-65%) --> wron
A1 10 8 2 ... -->correc
A2 ... ... ... ... --> correc
A3 ... ... ... .. --> correc
A4 0 0 0 NULL --> correc

The problem occures when ProdA for 2003 and 2004 are both zero. I am getting un-predicable results (the number changes from time to time) in the % fc column. But the individual product item is fine. I checked for NULLs and ZEROs before I do the division. What else can I do

I tried growth/lastyearthismonth and (thisYearthismonth - lastyearthismonth)/lastyearthismonth. Same results.

Based on the sample data shown below, are you using a
custom rollup formula for Prod_dim? In that case, Solve
Order of the % fc_var could be relevant - you can try
changing the Solve Order of that measure.

lastyearthismonth). I used the ParallelPeriod function. It
works ok.
I have ProdA and ProdB. Under ProdA, I have SubA; Under
SubA, I have A1, A2...
... --> correct
.. --> correct
changes from time to time) in the % fc column. But the
individual product item is fine. I checked for NULLs and
ZEROs before I do the division. What else can I do?