topic Re: Sum by two groups in SAS Programming

Sum by two groups

leonzheng — Wed, 02 May 2018 17:39:41 GMT

I have two datasets as below:

data2016:

item	quality	Jun	Jul	Aug
apple	bad	1	1	3
apple	good	6	8	9
orange	bad	1	2	1
orange	good	4	9	6
peach	bad	2	2	3
peach	good	10	12	17

data2017:

item	quality	Jun	Jul	Aug
apple	bad	2	2	3
apple	good	13	14	15
orange	bad	3	5	1
orange	good	8	7	13
cherry	bad	1	4	2
cherry	good	7	9	5

Now I need to make a summary result as below:

summary:

item quality Jun Jul Aug

item	quality	Jun	Jul	Aug
apple	bad	3	3	6
apple	good	19	22	24
orange	bad	4	7	2
orange	good	12	16	19
cherry	bad	1	4	2
cherry	good	7	9	5

peach	bad	2	2	3
peach	good	10	12	17

Logic is that sum the data of 2016 and 2017 result for each item, quality and month. I tried use cumulative sum and remove some rows but do not know how to calculate cumulative sum for several month like Jun Jul Aug, or is there any other easier way of doing this?

My cumulative sum code:

proc datasets;
append base=data2016 data=data2017;
run;

proc sort data=data2016;
by item quality;
run;

data summary;
set data2016;
by item quality;
if first.flag then June=Jun July=Jul August=Aug;
else June=Jun+June July=Jul+July August=Aug+August;
run;

then I will drop and rename columns and join summary having Aug=max(Aug), haven't done this step because the sum is still not working.

Please help to see how make the sum step working, and is there any other easier way to do this?

Thanks,

Re: Sum by two groups

PaigeMiller — Wed, 02 May 2018 17:42:14 GMT

Step 1: append the two data sets

Step 2: use PROC SUMMARY or PROC MEANS to do the sums, you need to set ITEM and QUALITY as CLASS variables.

Re: Sum by two groups

Reeza — Wed, 02 May 2018 17:42:57 GMT

Don't use a data step to calculate summary statistics unless you're doing really customized calculations.

Try PROC MEANS instead and include both Item and Quality in your BY or CLASS statements.

This shows you how to use PROC MEANS with multiple variables and save the output to a data set. You can replace my feature1-feature3 with your Jun-Aug. You can run the code in the example below to see how the data is structured and how the results are generated.

https://github.com/statgeek/SAS-Tutorials/blob/master/proc_means_basic.sas

Re: Sum by two groups

leonzheng — Wed, 02 May 2018 17:43:29 GMT

thanks, can you show me the code please?

Re: Sum by two groups

PaigeMiller — Wed, 02 May 2018 17:44:33 GMT

There are examples in the PROC MEANS documentation

http://documentation.sas.com/?cdcId=pgmmvacdc&cdcVersion=9.4&docsetId=proc&docsetTarget=p0f0fjpjeuco4gn1ri963f683mi4.htm&locale=en

Re: Sum by two groups

leonzheng — Wed, 02 May 2018 17:57:40 GMT

thans, I use

output out=summary sum=/ autoname;

then it works.

Re: Sum by two groups

leonzheng — Wed, 02 May 2018 17:57:50 GMT

thank you

Re: Sum by two groups

Reeza — Wed, 02 May 2018 18:32:45 GMT

Yes, or you can explicitly name them as desired. I usually try to have a common prefix since SAS works on prefixes not suffixes.

output out=want mean(jun)= Avg_Jun mean(jul)=Avg_Jul .. etc;