filtering row in output data set doesn't work

pkansal
Level 3
filtering row in output data set doesn't work

I am trying to put the following filter in Configure Sample setting page:

charged_phone_number == 31643000430

 

However, it doesn't return anything. When I download the dataset and put the filter locally, it does contain the value in that column.

 

My dataset lies in S3. Is it possible that filter records doesn't work in S3?

What is happening?

0 Kudos
5 Replies
tgb417

@pkansal 

Just a quick thought.  Check your data type.  You might want to force the column with the phone number in the schema to string.  I’ve seen similar things happen with US postal codes that start with 0.  DSS will sometime try to be overly helpful and convert things that look like numbers to numbers, and then matching does not work.

Or you already have the phone number as stings and your comparison is of a number.  You might also try putting quotes around your number in your comparison.  

just a few thought….  

--Tom
0 Kudos
pkansal
Level 3
Author

No, it is stored as an integer only. Even if it is an integer, it should work, right?

0 Kudos
tgb417

@pkansal ,

I've seen time where I've had to play around with the scheme to get certain things working.  Sometimes with these very large integers I've seen things converted to Scientific Notation and other such anomalies.

The other thing to do is to reach to the Dataiku Support team.  There really are great at what they do and can help you with the very specific challenges you are having.

--Tom

--Tom
0 Kudos
Jurre
Level 5

not necessarily @pkansal , all int's are not equal. The number you are filtering on is outside the 'standard' int32-range (max : 2,147,483,647). Datatype bigint would be the datatype of choice if you insist on having it stored as a number.  Personally i only use any kind of numbertype when values have a numerical meaning and use; Telephone numbers don't get summed up, they are just identifiers. 

Not sure if that's the underlying issue here but i concur with @tgb417  that it's worth checking.

Berilio
Level 2

Hi @pkansal , i agree with @Jurre, but maybe you can try with:   toNumber(charged_phone_number == 31643000430