topic Re: Identifying Outlier Value in New SAS User

Identifying Outlier Value

anissak1 — Sun, 02 Jun 2019 22:22:35 GMT

We are analyzing data in the following format. Our question is,

how can we identify IDs where the difference between any 2 sequential values for "Level" is more than 2?

We have looked through SAS resources and, as relatively new programmers, can't figure this out.

Thank you!!

ID Date Level

1 01/01/2000 1

1 01/02/2000

1 01/05/2000 3

1 01/05/2010

2 01/06/2010 5

2 02/12/2005 6

Re: Identifying Outlier Value

Reeza — Sun, 02 Jun 2019 23:03:17 GMT

How do you want to deal with cases where you're missing values between sequential values as in your data? Show your expected output please.

In general, you likely want the DIF function but it's a bit tricky to use if you're not familiar with SAS.

https://documentation.sas.com/?docsetId=lefunctionsref&docsetTarget=n10t6w4op574ttn11yk68dottp7f.htm&docsetVersion=9.4&locale=en

@anissak1 wrote:

We are analyzing data in the following format. Our question is,

how can we identify IDs where the difference between any 2 sequential values for "Level" is more than 2?

We have looked through SAS resources and, as relatively new programmers, can't figure this out.

Thank you!!

ID Date Level

1 01/01/2000 1

1 01/02/2000

1 01/05/2000 3

1 01/05/2010

2 01/06/2010 5

2 02/12/2005 6

Re: Identifying Outlier Value

anissak1 — Mon, 03 Jun 2019 22:31:54 GMT

Thank you for the note! We would like to ignore the observations for which we have no value.

Re: Identifying Outlier Value

Astounding — Tue, 04 Jun 2019 06:50:59 GMT

Did you mean greater than or equal to 2? That's an easy tweak to the logic if necessary.

Here's a way to identify all observations that increased by 2 or more:

data want;
set have;
by id;
where level > .;
if dif(level) > 1;
if first.id = 0 ;
run;

So there's still some work to do, but this gets you through the hard part. Make me proud and take it home the rest of the way.

Re: Identifying Outlier Value

anissak1 — Thu, 06 Jun 2019 14:37:58 GMT

Thank you!! I guess you are a master for a reason. 😊. Much appreciation. We are close to crossing the finish line