Sampling Question

[agimpel]

I am trying to sample a population with specific parameters and would love some assistance. I have a population of "references" that have various risk ratings of Low, Medium and High. Each reference is associated with a business process as well. I want to create a sampling workflow so I can sample 1/3 of each of the business process references. How would i do this? example table below 

Reference	Risk Rating	Business Process
1214589	Low	Accounts Payable
1214591	Low	Bank Reconciliations
1214741	Low	Bank Reconciliations
1214742	Medium	Cash
1214745	Medium	Pension
1215115	High	Pension
1215110	High	Accounts Payable
1215389	High	Accounts Payable

[Chaoued]

can you explain what's the expected outcome?

[agimpel]

I want to sample 1/3 of the "references" from each business process so for example 1/3 of the accounts payable references etc. 

[Chaoued]

that's the result you want ?

Reference	Risk Rating	Business Process
1214589	Low	Accounts Payable
1214591	Low	Bank Reconciliations
1214742	Medium	Cash
1214745	Medium	Pension

[OTrieger]

@agimpel 

One way to do it is using filter to get the different rick rate, then connect it to sample tool and set it to 30%, then union the data back.

[OTrieger]

You can do it also with a batch macro that the Control Parameter is the Risk Rating

[Pilsner · Accepted solution]

Hello @agimpel,

One way to do this is to use the sample tool to select "1 of every N rows". By setting N = 3 and selecting a grouping mechanism of Business Process, I believe you will get the result you are after.

I am unsure weather you need the answer to be random or not, but to enforce a random selection, you can assign a random value to each record, then sort ascending. That way, all the order will be random each time, therefore, the values sampled will also be random. 

I have attached the workflow below in case you will find it helpful. Please let me know how you get on.

Regards - Pilsner