TM1 Forum

Hi,

I've been struggling with the fact that the decimal separator in my data is "," and there are no thousand separators. If I try defining these settings in the data source tab of Performance Modeler, the numbers imported are way too big, in other words, TM1 seems to read the commas as thousand separators.

However, as soon as I defined DatasourceASCIIThousandSeparator=''; and DatasourceASCIIDecimalSeparator=','; on the prolog tab, I get an error saying "Cannot convert value ... to a real number" (Cannot_convert.PNG) and no values are imported. In the mapping tab data preview the data looks right (decimal.PNG) and the data type is set to decimal. Moreover, I cannot import the data values straight from the source file (field number 3 in the error log is the original value) and I have to use a calculated column that switches the sign of the data while needed (this cannot apparently be done by weights, since the data needs to be seen with a different sign also on the leaf level) To sum it up, in Performance Modeler I've only managed to import too big values or no values at all.

I've tried changing language settings and was able to import correct numbers with Architect, but I'd rather use Performance Modeler since I have no experience of Architect. Anyhow, while changing the language settings, the user interface language changes (as do the names of the variables) and I'd rather use Performance Modeler in English, to be able to search error messages in the future.

Could someone tell what is the best practice if I want to use Performance Modeler in English, but to have comma as a decimal separator and no thousand separator at all? I would be really grateful for any answers!

Not played with this much but:

marine wrote: However, as soon as I defined DatasourceASCIIThousandSeparator=','; and DatasourceASCIIDecimalSeparator=''; on the prolog tab,

That sounds the wrong way round; your data source looks to have:

declanr wrote:Not played with this much but:

marine wrote: However, as soon as I defined DatasourceASCIIThousandSeparator=','; and DatasourceASCIIDecimalSeparator=''; on the prolog tab,

That sounds the wrong way round; your data source looks to have:

Code: Select all

DatasourceASCIIThousandSeparator='.';
DataSourceASCIIDecimalSeparator=',';

Hi, my bad, I was supposed to write DatasourceASCIIThousandSeparator=''; and DatasourceASCIIDecimalSeparator=',';

So I tried changing the decimal separator to comma, only put it the wrong way around in my message.

I have had a similar issue before where between TM1 and the OS there was confusion between the Decimal Symbol, Digit Grouping Symbol and List Separator. Try changing these in the OS so that they do not conflict with what you are importing.

Run INTL.CPL from the Run box or command prompt then choose additional settings to do this. If this is on a server, you will need to update for the user who starts the TM1 Service for the TM1 Model/System.

gtonkin wrote:I have had a similar issue before where between TM1 and the OS there was confusion between the Decimal Symbol, Digit Grouping Symbol and List Separator. Try changing these in the OS so that they do not conflict with what you are importing.

Run INTL.CPL from the Run box or command prompt then choose additional settings to do this. If this is on a server, you will need to update for the user who starts the TM1 Service for the TM1 Model/System.

Thank you for your answer! I was just wondering which settings would affect the Decimal Symbol, Digit Grouping Symbol and List Separator, without affecting the language of the TM1 user interface? Is it possible to change the language of Performance Modeler separately, or does it always depend on my Windows settings? If I remember correctly, changing Format to Finnish changed also the language settings in PM.

I would try changing the List Separator, Digit Grouping etc. under additional settings without changing other language related settings.
Not sure on the Performance Modeler settings - you will have to play around and test various scenarios.

gtonkin wrote:I would try changing the List Separator, Digit Grouping etc. under additional settings without changing other language related settings.
Not sure on the Performance Modeler settings - you will have to play around and test various scenarios.

Thanks for the advice! In case someone else is struggling with the issue, I found a workaround for this in Architect: http://www-01.ibm.com/support/docview.w ... wg1PM78147.

In Performance modeler, for some reason, when using Guided Import, it is enough to change the settings on data tab.