Computing coefficient of variation in two dimenstional groups

[Sejin]

Hello.

I am having a trouble in computing cv in two dimensional groups.

My dataset looks like this.

class1	class2	id	var1	var2
1	x	a	4	5
1	y	a	4	5
2	x	b	3	6
2	x	c	.	3
3	x	a	4	5
3	x	b	3	6
3	y	c	.	3

I want to calculate cv of var1 and var2 for each class1-class2 groups such as 1-x, 1-y, 2-x, 3-x, 3-y.

I also want to skip missing variables in computing cv.

I tried

proc sql;

create table want as

select *, std(var1) as std_var1, mean(var1) as mean_var1, calculated std_var1/calculated mean_var1 as cv_var1

from have

group by class1, class2;

quit;

but I couldn't get the data that I want.

If you have any idea, plz share it with me.

Thanks for reading the thread.

[PaigeMiller]

---

@Sejin wrote:

Hello.

I am having a trouble in computing cv in two dimensional groups.

My dataset looks like this.

class1	class2	id	var1	var2
1	x	a	4	5
1	y	a	4	5
2	x	b	3	6
2	x	c	.	3
3	x	a	4	5
3	x	b	3	6
3	y	c	.	3

I want to calculate cv of var1 and var2 for each class1-class2 groups such as 1-x, 1-y, 2-x, 3-x, 3-y.

I also want to skip missing variables in computing cv.

I tried

proc sql;

create table want as

select *, std(var1) as std_var1, mean(var1) as mean_var1, calculated std_var1/calculated mean_var1 as cv_var1

from have

group by class1, class2;

quit;

but I couldn't get the data that I want.

---

Be specific. What do you get (show us), and what do you want (show us).

[Sejin]

When I ran my code, I got mean values by groups just as I wanted, but I could not pass missing values in calculation.

Also, in calculating standard deviation values, it neither calculated the value group by group nor passed missing values.

[PaigeMiller · Accepted solution]

Show us what you get. Show us what you want.

[Sejin]

Well.. it suddenly worked

srry for waisting your time

[Ksharp]

Why not use 

proc means data=have cv ;

by class1 class2;

var var1 var2;

run;