topic Re: Mean imputation in SAS Programming

Mean imputation

Naveen1 — Mon, 09 Oct 2017 11:33:15 GMT

Hi,

Need help, I want to impue the data for grouped data based on variabke Region,

Data set:

Country	Region	Age
Nigeria	Africa	15
Ethiopia	Africa
Egypt	Africa	42
Democratic Republic of Congo	Africa
South Africa	Africa	13
Russia	Europe	45
Germany	Europe	54
France	Europe	17
United Kingdom	Europe	60
Italy	Europe

Africa Region mean : 23 and Europe Region mean :44

Expected Output:

Country	Region	Age
Nigeria	Africa	15
Ethiopia	Africa	23
Egypt	Africa	42
Democratic Republic of Congo	Africa	23
South Africa	Africa	13
Russia	Europe	45
Germany	Europe	54
France	Europe	17
United Kingdom	Europe	60
Italy	Europe	44

I have so many Levels in (Regions), by writing if condition will take much time. So is there any other method ...?

Missing value Imputation for Grouped Data

Naveen1 — Mon, 09 Oct 2017 11:10:13 GMT

Hi,

Need help to impute the missing values by mean for the group data,

Data :

Country	Region	Age
Nigeria	Africa	15
Ethiopia	Africa
Egypt	Africa	42
Democratic Republic of Congo	Africa
South Africa	Africa	13
Russia	Europe	45
Germany	Europe	54
France	Europe	17
United Kingdom	Europe	60
Italy	Europe

Mean of Africa=23 and Mean of Europe =44

Expected Output:

Country	Region	Age
Nigeria	Africa	15
Ethiopia	Africa	23
Egypt	Africa	42
Democratic Republic of Congo	Africa	23
South Africa	Africa	13
Russia	Europe	45
Germany	Europe	54
France	Europe	17
United Kingdom	Europe	60
Italy	Europe	44

Re: Mean imputation

gamotte — Mon, 09 Oct 2017 12:09:46 GMT

Hello,

With SQL :

proc sql;
  CREATE TABLE want AS
  SELECT country, region,
         CASE WHEN age NOT IS MISSING THEN age ELSE round(mean(age)) END AS age
  FROM HAVE
  GROUP BY region;
quit;

Re: Missing value Imputation for Grouped Data

Ksharp — Mon, 09 Oct 2017 12:29:44 GMT

data have;
infile cards expandtabs truncover;
input Country	& $40. Region & $20.	Age;
cards;
Nigeria 	Africa	     15
Ethiopia	Africa	 
Egypt	Africa	  42
Democratic Republic of Congo	Africa	 
South Africa	Africa	 13
Russia	 Europe	 45
Germany 	Europe	     54
France	 Europe	 17
United Kingdom	 Europe	 60
Italy	Europe	
;
run;
proc stdize data=have out=want reponly missing=mean;
by region;
var age;
run;

Re: Mean imputation

Reeza — Mon, 09 Oct 2017 13:27:30 GMT

Look at PROC STDIZE

Re: Mean imputation

andreas_lds — Mon, 09 Oct 2017 15:38:44 GMT

And another double-post: https://communities.sas.com/t5/Base-SAS-Programming/Missing-value-Imputation-for-Grouped-Data/m-p/402316

Please don't post one question in multiple groups, this just causes confusion and those willing to help may waste time.

Re: Mean imputation

Reeza — Mon, 09 Oct 2017 18:10:59 GMT

@andreas_lds @Naveen1 please note that I've merged your duplicate posts into a single one.