Data Precision - Transformer

[spicysudhi]

Hi

in the iqd file, i for few numeric columns. When i view the data from the model in transformer, only one column data is devided by 100 (if valus is 7.4, i get it as 0.074). all other columns are fine. I managed to get rid of this problem by setting the precision value for the column in datasource. question is why this happening.
thanks
sudhi

 

[sln007]

Hi,
Not sure what you are asking.
Sometimes measures seem to 'inherit' the formatting attributes of other measures in the list. Just setting the
format method for measures will fix that.
I haven't experienced issues related to the datasource columns being formatted improperly.
If you supply more details maybe I could shed some light on your issue.

 

[spicysudhi]

data is like this (in table)

col1 col2 col3
10.4 12.4 7.4

in transformer, in data source view data loooks like this
col1 col2 col3
10.4 12.4 0.074

in the cube data looks like same as it seen in data source viewer.

(see the value of col3.)

i changed the col3 properties in data source by setting precision to 2 and solved this problem, but why this is happening, no answer with me.

regards,
sudhi

 

[GJParker]

I've never quite sorted this one out but a couple of things to look at

are the original data types different ?
are the measure storage types different ?


Gary Parker
Systems Support Analyst
Manchester, England

 

[spicysudhi]

original data type number, same size.

measure types...??

 

[GJParker]

Measure properties, bottom left hand corner Storage Type

16-bit integer
32-bit integer
64-bit floating point

Gary Parker
Systems Support Analyst
Manchester, England

 

[spicysudhi]

64-bit integer

 

[sln007]

8645,

It's just always a good idea to set the "format" attributes
of your measures in the model for a cube.
It's a simple solution to a minor issue.

 

[DoubleD]

This is a bug within Impromptu. When reading certain numeric fields, Cognos improperly interprets the datatype. To correct this, change col3 to "col3 * 1" in your IQD file. Now Transformer will correctly interpret this field.

This seems like a silly fix, but it's what Cognos suggests and it's the best way I've found to deal with it. Messing with precisions in the Transformer model can bite you if Cognos corrects the datatype issue, or if the datatype ever changes (i.e. Number 3,2 to Number 3,4).

Pain is stress leaving the body

DoubleD