Skip to main content

Hi everyone, 

Just wondering if Ataccama provides any best practice guides on naming conventions when working with the platform. This includes best practice for rule names, error names, sql catalogue item names etc. 

 

Kind Regards, 

 

Vishy

Hi @vdatla ,

Here are some best practices while creating any rules  The rule must have a descriptive name. Since this is a collaborative platform, we want others to
understand what the rule does and reuse it, instead of re-creating the same rule from scratch.
The rule name should have the format <attribute> is/has/contains/etc <missing, invalid,
duplicate, between 0 and 10,000 etc>. Examples for rule names include:
● Account ID is all zeros
● Attribute (string) is duplicate
● Attribute (string) is missing
● Attribute (date) is missing
● Email address is test
● Email address format is invalid
● ID contains a lowercase character
● Market code is invalid
● Account Balance is between 0 and 10,000

 

 

Hope this helps !

Regards,

Srija Piratla


Hi!

When it comes to ONE desktop, we have a pretty standard best practice list for column prefixes:

src_  : The source value of that field, it does not get altered

pur_ : This is a column that had transformations done to it

std_ : This is a column that has been standardized, usually with a dictionary or standardization transformation

cio_ : The best possible value that can be achieved after processing is done, usually the std_

mat_ : This is used for matching purposes only, for MDM for example

dic_ : A value that was returned or verified by using a dictionary

 

When it comes to scoring, we want to score between 0 (no changes or transformations done) to 10,000,000 (null value) for naming these scoring reasons, we can put things like “ID_IS_NULL” and set the score to 10,000,000 , or we can do “ADDRESS_INCORRECT” and set the score to 1,000,000. 

Usually when scoring, we will set a null value to be the score of 10,000,000 , while ones that have smaller rules broken to be under 10,000,000

 

Some more tips:

  • Try to be as descriptive as possible, but keep the names within a reasonable length
  • “customers_NA” vs “customers_north_america” the customers_NA does the job and is more compact
  • “FIRST_NAME_IS_NULL” vs “FN_NULL” we can use the first letter of words to get the same point accross and keep it more compact
  • There are times though where we can also just shorten the word a bit like “FST_NM_NULL” it's nice and compact but does the job! it helps describe better what it is while also not letting it get too long

 

Hope this helps! -

Samuel :) 

 


Hi @vdatla, welcome to the community and thank you for posting! I’m closing this thread for now, if you have any follow-up questions please do share them in the comments below or create a new post 🙋‍♀️


Reply


ataccama
arrows
Lead your team  forward  OCT 24 / 9AM ET
×