Precision values tm1

[ravi]

Hi

I have a issue with my Active report where the Gross margin numbers for a couple of lines show huge numbers due to division by a number close to 0
this may be because of the Precision values.
Please advise me how to reset Precision values in tm1?
ravi

[qml]

There is no way of influencing TM1's calculation precision, it's only possible to control the displayed format/precision.

It has been discussed recently, namely here: http://www.tm1forum.com/viewtopic.php?f=3&t=6072.

[ravi]

Hi qml,

This issue is only for few of the lines in my active form- Calculatiion ade decimal precition looks good for so many cells.

you said- it's only possible to control the displayed format/precision.could you please give me steps to do this?
Thank you
ravi

[qml]

Your issue, whatever it is (because I don't feel like you've really described it well), does not sound like a formatting issue. TM1, when calculating, uses the whole available precision. And if you divide by a number very close to 0 you're bound to have a result that is very large as well. This is pure mathematics. Why would you expect any differently?

However, to answer your question on formatting, in Cube Viewer you set the format by using menu Options -> Format.... You can also set up default formats for dimension elements using the Format attribute of each dimension. Finally, in Excel you can use Excel formatting by choosing Format Cells... from the context menu. In the case of Active Forms you need to define that formatting in the hidden Format Area at the top of the sheet.

Edit: If you want to influence the precision used during calculations and make TM1 treat, say, 0.00000000001 as 0 then you can look at using the ROUNDP() function in your rules or TI.
Example: ROUNDP(0.0001, 3) will return a value of 0, whereas ROUNDP(0.0001, 4) will return 0.0001.

[moby91]

See the IBM Technote 1515228:


https://www-304.ibm.com/support/docview ... wg21515228
TM1 Precision

Technote (FAQ)

Question
What is the decimal precision of TM1?

Answer
TM1 stores data in IEEE-754 Floating Point Double.
The maximum number of digits is 15+ (almost 16, but some 16 digit number will not fit with full precision).
TM1 will clip to 14 digits for display.
By default Architect will display 9 digits of precision after the decimal point. This can be modified by by defining a custom format for the cell. Of course, the same overall 15 digit limit still applies. If one has 10 digits to the left of the decimal point, there will be only 5 digits to the right.

[ravi]

Hi qml,

I am not too sure to resolve this issue at a cube level.
are you saying that I have to write TI process/rules to resolve this issue?

attached screenshot-where are the values are calulcated by dividing 0(zero)

what could be the IT process or a rule- please advise.
Regards,
Ravi

[qml]

are you saying that I have to write TI process/rules to resolve this issue?

Surely these lovely numbers you have shared with us are a result of some calculation, which is either a rule, or a TI, right? I'm hoping you are able to find that calculation and then spend some time thinking what is actually happening and how you can apply the suggestions from the above posts to that calculation to alleviate/solve that particular problem you're facing.