Converting Numeric field to Character field without losing the precision

[sbrighenti]

Hello,

I am currently working in DiStudio and am reading in a SAS dataset that has a numeric value. I'm trying to create a 'verification' job that looks to see the number of digits before and after the decimal and flag any that are not accurate. I've done this before, but the field was a character field and not numeric.

For example, we have a table of values and the ones that appear in a numeric column are as follows:

1.1230

1.784

1.1487

We want to find all decimals without four decimals (ie 1.784) However, when I do (LENGTH(SCAN(PUT(X,best.),2,'.')) ^= 4, it flags both 1.784 and 1.1230 since best drops the trailing zero.

Would somebody be able to help me figure out how I can find the digits after the decimal place in the WHERE tab of DiStudio?

[SuryaKiran]

Don't convert to numeric, extract the number before and after decimal and find the length for that characters.

length(strip(scan(val,1,'.')))=1 and  length(strip(scan(val,2,'.')))=3

[SuryaKiran]

Also you can try perl regular expression for this type of matching.

prxmatch("/^\d{1}\.\d{3}$/",strip(val))=1

^           - Beginning 

\d{1}     - one digit

\. (\dot)  - escape special character { . } 

\d{3}      - 3 digits 

$           - End of line

[Patrick]

Once you've read the source string into a numerical variable there is no telling if the source string was 1.2 or 1.2000

It's numerically the same and will be stored internally the same. It has nothing to do with formats.

To check the digits of the source string you need to read the data as character. With DIS you can do so by defining a character informat for the External File Metadata. After that use the same approach as in the past. You can always later on convert the character variable to a numeric one via an expression like input(<charvar>, best32.)