proc means (where and in operator).

[xoxozav_1]

Hi,

I am struggling to generate different groups in proc means. I am trying to generate weekly average temperature groups using id variable and IN operator, and then get their average of each group.

I developed the following syntax, but something is wrong. Could someone help me find what the mistake is?

thank you.

This is my syntax :

proc means data= id80_00_20a;
where id (in (1,2,3,4,5,6, 7))=1;
where id (in (8,9,10,11,12,13,14, 15))=2 ;
where id (in (16,17,18,19,20,21,22,23))=3;
where id (in (24,25,26,27,28,29,30, 31))=4;
class id;
var avg1 avg2 avg3 ;
run;

This is what the log page says:

1 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;

72

73 proc means data= id80_00_20a;

74 where id (in (1,2,3,4,5,6, 7))=1;

ERROR: Function IN requires a character expression as argument 1.

75 where id (in (8,9,10,11,12,13,14, 15))=2 ;

ERROR: Function IN requires a character expression as argument 1.

76 where id (in (16,17,18,19,20,21,22,23))=3;

ERROR: Function IN requires a character expression as argument 1.

77 where id (in (24,25,26,27,28,29,30, 31))=4;

ERROR: Function IN requires a character expression as argument 1.

78 class id;

79 var avg1 avg2 avg3 ;

80 run;

[Tom · Accepted solution]

The syntax for IN is 

value in (...list...)

and it returns a boolean results.

Why not make a format?

proc format ;
value idgroup
 1-7='Group 1'
 8-15='Group 2'
16-23='Group 3'
24-31='Group 4'
;
run;

proc means data= id80_00_20a;
  class id;
  var avg1 avg2 avg3 ;
  format id idgrp.;
run;

[ChrisNZ]

1.  A where clause is used to subset a table.

Here you have 4 clauses, so the last one supersedes the others.

2.  It seems that you want to change the value of ID.

A where clause does not do this.

3. The syntax you use is invalid. This syntax is valid:

where ID in (1,2,3,4,5,6,7) ;

[ballardw]

---

@xoxozav_1 wrote:

Hi,

I am struggling to generate different groups in proc means. I am trying to generate weekly average temperature groups using id variable and IN operator, and then get their average of each group.

I developed the following syntax, but something is wrong. Could someone help me find what the mistake is?

thank you.

This is my syntax :

proc means data= id80_00_20a;
where id (in (1,2,3,4,5,6, 7))=1;
where id (in (8,9,10,11,12,13,14, 15))=2 ;
where id (in (16,17,18,19,20,21,22,23))=3;
where id (in (24,25,26,27,28,29,30, 31))=4;
class id;
var avg1 avg2 avg3 ;
run;

This is what the log page says:

1 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;

72

73 proc means data= id80_00_20a;

74 where id (in (1,2,3,4,5,6, 7))=1;

ERROR: Function IN requires a character expression as argument 1.

75 where id (in (8,9,10,11,12,13,14, 15))=2 ;

ERROR: Function IN requires a character expression as argument 1.

76 where id (in (16,17,18,19,20,21,22,23))=3;

ERROR: Function IN requires a character expression as argument 1.

77 where id (in (24,25,26,27,28,29,30, 31))=4;

ERROR: Function IN requires a character expression as argument 1.

78 class id;

79 var avg1 avg2 avg3 ;

80 run;

---

If by any chance you had an actual date value for each record you could use a FORMAT to get values by week, depending on your "week" definiton. SAS uses 3 different appearances look at WEEKU WEEKV and WEEKW for the differences in starting day of the week and cross year behavior. This shows data that crosses a year boundary with a 2 digit year and the week indicator.

data have;
   do date='14Jul2018'd to '05Mar2019'd;
      randomvalue = 100*(rand('uniform'));
      output;
   end;
run;

proc summary data=have nway ;
   class date;
   format date weekv5.;
   var randomvalue;
   output out=work.summary mean=;
run;

You could create a nicer format for appearance using Proc Format.

An advantage of the DATE plus format is that you can create other summary groups just changing the format. YYMM would give a year/month summary, YYQ would give a year/quarter summary, YEAR would give a calendar year summary, MONTH would give a calendar month summary across years and more with other formats.

Or starting with a date you could create custome intervals with the INTNX function to create a date representative of LOTS of things.