topic Re: The extended penalty problem... in SAS Procedures

The extended penalty problem...

sweetgorgor — Wed, 20 Nov 2013 22:49:36 GMT

Hi, first let me apologize for not being more clear on the title as I am not sure how to discribe what I am looking for in just a few words. I am essentially trying to derive the Extended Penalty column in the table below:

Date	Mistakes_Flg	Number Of consecutive mistakes	Penalty	Extended_Penalty
1/1/2013	0	0	0	0
1/2/2013	1	1	10	0
1/3/2013	1	2	10	0
1/4/2013	0	0	0	0
1/5/2013	1	1	10	0
1/6/2013	1	2	10	0
1/7/2013	1	3	10	0
1/8/2013	1	4	10	0
1/9/2013	0	0	0	10
1/10/2013	0	0	0	10
1/11/2013	0	0	0	10
1/12/2013	0	0	0	10
1/13/2013	0	0	0	0

where if the person makes mistakes for 3 or more consecutive days, he gets punished by paying an daily extended penalty of the previous penalty equal to the number of consecutive days he made a mistake. In my example, the first time he made a mistake for 2 consecutive days, no extended penalty has incurred since it's less than the 3 days criteria. However, he then made a mistake for 4 consecutive days, therefore he will have to pay the extended penalty for 4 more days after he stop making mistakes. Can someone help I tried everything and I can't get it work!

Thanks a bunch in advance!

Re: The extended penalty problem...

Reeza — Wed, 20 Nov 2013 22:56:53 GMT

What do you have working? I'm assuming you have an ID field somewhere as well?

If someone makes a mistake for 3 days then there's a 3 day penalty or is more than 3?

What happens if they make mistakes for 8 days, stop for 4 days and then starts again?

Re: The extended penalty problem...

SteveDenham — Thu, 21 Nov 2013 14:34:36 GMT

Or extending Reeza's question:

-Mistakes for 8 days, followed by no mistakes 4 days --> Extended penalty on the 4 no mistake days

-This is then followed by: Mistakes for 3 days after no mistakes for 4 days, followed by no mistakes for 4 days --> Extended penalty on the 3 mistake days. I'm clear to this point. Now what happens on the first no mistake day in this section--is it double extended penalty (1 day carried over from the 8 day period, plus the first from the more recent 3 day period), or just extend from the recent 3 day and have a single extended penalty?

This is a lot like coding for disease cases. If we say that no new case can begin until after n days have passed since the last day of signs or symptoms the logic is similar. It is just that in the epidemiology situation n is usually predefined, while here it is a matter of looking backward.

Steve Denham

Re: The extended penalty problem...

sweetgorgor — Thu, 21 Nov 2013 15:36:56 GMT

you guys are so awesome! I was comtemplating the same issues while looking into the contracts (People who wrote those contracts never thought of those issues!) I will try to simply the problem first, if i still have some questions in regards to the coding aspect I will follow it up on this thread. Thanks!

Re: The extended penalty problem...

snoopy369 — Thu, 21 Nov 2013 16:00:04 GMT

To answer the 'simple' case, I think the easiest method to explain is simply to have one more column: "Number of remaining extended penalty days". Make it work exactly like your "Number of consecutive mistakes" column, but in reverse - count down from 4 to 3 to 2 to 1.

Re: The extended penalty problem...

Aad — Fri, 22 Nov 2013 14:10:13 GMT

The excellent explanation of your problem challenged me to write a 9-statement solution plus comments as follows. Hope you enjoy it ...

Data Mistakes;

/*-----------------------------------------------------------------*/

/* Generate simulation table with dates and some random Mistakes */

/*-----------------------------------------------------------------*/

Attrib Date Length = 4 Format = Date11.

Mistake Length = 2 Format = 1.;

Do Date = '01-jan-2013'd to '31-dec-2013'd;

Mistake = Round (RanUni (5), 1);

Output;

End;

Run;

/*1*/Data Mistakes;

/*-----------------------------------------------------------------*/

/* Re-generate table, including PENALTIES_TO_GO and PENALTY values */

/*-----------------------------------------------------------------*/

/*2*/ Attrib Date Length = 4 Format = Date9.

Mistake Length = 3 Format = 1.

Penalty Length = 3 Format = 1.

Penalties_to_go Length = 3 Format = 8.

/*3*/ Set Mistakes;

/*-----------------------------------------------------------------*/

/* PENALTY becomes 1 if current and last 3 dates all had MISTAKE=1 */

/*-----------------------------------------------------------------*/

/*4*/ Penalty = Mistake * Lag1 (Mistake)

* Lag2 (Mistake) * Lag3 (Mistake);

/*-----------------------------------------------------------------*/

/* Accumulate the number of PENALTIES_TO_GO */

/*-----------------------------------------------------------------*/

/*5*/ Penalties_to_go + Penalty;

/*-----------------------------------------------------------------*/

/* If current+last 3 days had MISTAKE=1, and 4 days ago MISTAKE=0, */

/* then increase the PENALTIES_TO_GO by an additional 3 penalties. */

/*-----------------------------------------------------------------*/

/*6*/ If Penalty > Lag4 (Mistake) >= 0 Then Penalties_to_go + 3;

/*-----------------------------------------------------------------*/

/* Assign a PENALTY if current date has MISTAKE=0 and there are */

/* still one or more PENALTIES_TO_GO. */

/*-----------------------------------------------------------------*/

/*7*/ Penalty = (Error = 0 and Penalties_to_go > 0);

/*-----------------------------------------------------------------*/

/* Decrease the number of PENALTIES_TO_GO only if PENALTY=1 */

/*-----------------------------------------------------------------*/

/*8*/ Penalties_to_go + -Penalty;

/*9*/ Run;

Proc Print Uniform;

Run;