PROC FORMAT - Categorizing a continuous measure

[_maldini_]

How do I capture all values of a continuous measure using formats?

For example: yrs_monthly_use = 5.3 (5.3 years of monthly use)

VALUE 
	yrs_monthly_usefmt
			low-<5 = "< 5 yrs"
			5-10 = "5-10 yrs"
			10-high = "> 10 yrs";

With the syntax above, 5.3 will end up in the 2nd category. Great.

Where does 10 end up? Also, in the 2nd category?

How do I modify the syntax so that the third category captures any value greater than 10?

I tried the following, but the label for the 2nd category becomes "5-10 yrs >":

	yrs_monthly_usefmt
			low-<5 = "< 5 yrs"
			5-10 = "5-10 yrs"
			>10-high = "> 10 yrs";

Thanks.

[Reeza · Accepted solution]

The extra > comes from the ">10-high" portion, not the specification of the range. 

You have the somewhat right idea, but you need to include the <  symbol in between the values and it means include the end points. 

proc format;
values	yrs_monthly_usefmt
			low-<5 = "< 5 yrs"
			5-<10 = "5-10 yrs"
			10-high = "> 10 yrs";
			
run;

data demo;
do i=0 to 12 by 0.5;
month = put(i, yrs_monthly_usefmt.);
output;
end;
run;

proc print data=demo;
run;

I believe by default the end point is included in the interval. But you can modify that with the < symbol. Try playing around with including it inside the ranges. 

This is a good one to test for example:

proc format;
value	yrs_monthly_usefmt
			low-<5 = "< 5 yrs"
			5<-<10 = "5-10 yrs"
			10-high = "> 10 yrs";
			
run;

5 doesn't get included in any range here, so it doesn't get formatted at all 🙂

---

@_maldini_ wrote:

How do I capture all values of a continuous measure using formats?

For example: yrs_monthly_use = 5.3 (5.3 years of monthly use)

VALUE 
	yrs_monthly_usefmt
			low-<5 = "< 5 yrs"
			5-10 = "5-10 yrs"
			10-high = "> 10 yrs";

With the syntax above, 5.3 will end up in the 2nd category. Great.

Where does 10 end up? Also, in the 2nd category?

How do I modify the syntax so that the third category captures any value greater than 10?

I tried the following, but the label for the 2nd category becomes "5-10 yrs >":

	yrs_monthly_usefmt
			low-<5 = "< 5 yrs"
			5-10 = "5-10 yrs"
			>10-high = "> 10 yrs";

Thanks.

---

Spoiler

This is the format you want most likely:



proc format;
value yrs_monthly_usefmt
low-<5 = "< 5 yrs"
5-10 = "5-10 yrs"
10<-high = "> 10 yrs";

run;

[Tom]

It is not hard to run a little test to see what it does.

1733  data _null_;
1734    do yrs=1,5,7,10,12 ;
1735      put yrs= yrs yrs.;
1736    end;
1737  run;

yrs=1 < 5 yrs
yrs=5 5-10 yrs
yrs=7 5-10 yrs
yrs=10 5-10 yrs
yrs=12 > 10 yrs

If you want to exclude one of the boundaries from a range place a < on that side of the hyphen.

proc format;
  VALUE yrs
    low-<5 = "< 5 yrs"
    5-<10 = "5-10 yrs"
    10-high = "10+ yrs"
  ;
run;

Result

1745  data _null_;
1746    do yrs=1,5,7,10,12 ;
1747      put yrs= yrs yrs.;
1748    end;
1749  run;

yrs=1 < 5 yrs
yrs=5 5-10 yrs
yrs=7 5-10 yrs
yrs=10 10+ yrs
yrs=12 10+ yrs