How can I avoid dataiku from changing the data types?

mlv05
Level 1
How can I avoid dataiku from changing the data types?
Hi,

I have a dataset with a field that it's supposed to be a string, the field contains four numbers but sometimes the first number is a cero (like 0999) and dataiku keeps reeding them as numbers even when I declare them as text.

The biggest problem is that when I try to export the dataset, the first cero dissapears. How can I stop this?
0 Kudos
7 Replies
Clรฉment_Stenac
Hi, it is not possible to prevent that if no zero appears in the sample. If a zero-leading number appears in the sample (other than a decimal), DSS should consider it as text automatically.

For some recipes, DSS will automatically propagate the forced "Text"

However, there is a catch with Excel export: Excel export does not currently respect the types, and will always remove the leading zeros. To work around, use CSV export instead of Excel export. Beware that when importing the resulting CSV into Excel, Excel will also remove the leading zero, unless you force "text" format in the Excel import wizard.
0 Kudos
mlv05
Level 1
Author
Thank you for your answer. I'm actually using CSV export but it still removes the firts zero
0 Kudos
Clรฉment_Stenac

Could you verify this by opening the CSV file with a text editor like notepad, rather than Excel?

0 Kudos
joostjansenn
Level 1

Is the same true for the scoring recipe of a model? It seems to remove leading 0s for me in strings as well @Clรฉment_Stenac 

0 Kudos
adiangels
Level 1

Facing same problem where the zeroes are trimmed when data like 01 are used in the formula. Any known work around for this ? 

0 Kudos
Jurre
Level 5

Hi @adiangels , @joostjansenn and @mlv05 ,

My workaround for this is adding a prefix-letter (z for example) to values which possibly contain a leading zero, before exporting. Nothing gets stripped then. 

Hope this helps!

Jurre

0 Kudos
Diwei
Level 2

HI Hi @adiangels , @joostjansenn and @mlv05 @Jurre ,

I also do the same, add ' in front of numbers. Another alternative is to write in Excel directly into a folder instead of dataset, then all the types remain intact. 

0 Kudos