topic Re: how to loops in following condition in SAS Programming

how to loops in following condition

K_coder — Mon, 04 Feb 2019 04:52:27 GMT

I have following data1(abc) which are representing coordinates-

data abc;
input index Cor1 Cor2 Cor3 Cor4 Cor5 Cor6 Cor7 Cor8;
datalines;
1 1 2 4 12 6 10 5 11
2 1 1.5 4 11.5 6.5 10.5 5 13
3 1.5 2 4 12 6 10 5 11
;

and data2(abc1) which represents classes w.r.t abc(data)

data abc1;
input class COORDINATE$;
datalines;
0 Cor1
0 Cor2
1 Cor3
2 Cor4
1 Cor5
2 Cor6 
1 Cor7
2 Cor8
;

For following code can we able to do it automatic, without writing series statement each time?

proc sgplot data=abc;
series X=index y=Cor1/lineattrs=(color=g);
series X=index y=Cor2/lineattrs=(color=g);
series X=index y=Cor3/lineattrs=(color=b);
series X=index y=Cor5/lineattrs=(color=b);
series X=index y=Cor7/lineattrs=(color=b);
series X=index y=Cor4/lineattrs=(color=r);
series X=index y=Cor6/lineattrs=(color=r);
series X=index y=Cor8/lineattrs=(color=r);
run;

if yes how?

Note:-

1. Maximum(class)=3

2. If classes increase it should be increase in colors for line.

Re: how to loops in following condition

Reeza — Mon, 04 Feb 2019 05:01:13 GMT

For following code can we able to do it automatic, without writing series statement each time?

Not with your current data structure. Change it so that CORR is one value and then the correlation ID (1 to 😎 is in another column:

Index ID Corr_Value

1 1 2

1 2 4

1 3 12

....

3 1 1.5

....

3 8 11

;

Then when you make your graph you can use the GROUP = to get separate lines:

proc sgplot data=abc;
series x=index y=corr_value / group = ID;

run;

You can control the colours with your separate data set, using a data attribute map.

@K_coder wrote:

I have following data1(abc) which are representing coordinates-
data abc;
input index Cor1 Cor2 Cor3 Cor4 Cor5 Cor6 Cor7 Cor8;
datalines;
1 1 2 4 12 6 10 5 11
2 1 1.5 4 11.5 6.5 10.5 5 13
3 1.5 2 4 12 6 10 5 11
;
and data2(abc1) which represents classes w.r.t abc(data)
data abc1;
input class COORDINATE$;
datalines;
0 Cor1
0 Cor2
1 Cor3
2 Cor4
1 Cor5
2 Cor6 
1 Cor7
2 Cor8
;
For following code can we able to do it automatic, without writing series statement each time?
proc sgplot data=abc;
series X=index y=Cor1/lineattrs=(color=g);
series X=index y=Cor2/lineattrs=(color=g);
series X=index y=Cor3/lineattrs=(color=b);
series X=index y=Cor5/lineattrs=(color=b);
series X=index y=Cor7/lineattrs=(color=b);
series X=index y=Cor4/lineattrs=(color=r);
series X=index y=Cor6/lineattrs=(color=r);
series X=index y=Cor8/lineattrs=(color=r);
run;
if yes how?

Note:-

1. Maximum(class)=3

2. If classes increase it should be increase in colors for line.

Re: how to loops in following condition

K_coder — Mon, 04 Feb 2019 05:30:50 GMT

How to create such data transformation?

Re: how to loops in following condition

Reeza — Mon, 04 Feb 2019 05:31:38 GMT

PROC TRANSPOSE
https://stats.idre.ucla.edu/sas/modules/how-to-reshape-data-wide-to-long-using-proc-transpose/

And an example of attribute maps:
https://blogs.sas.com/content/graphicallyspeaking/2013/04/02/attribute-maps-1/

Re: how to loops in following condition

K_coder — Mon, 04 Feb 2019 06:19:00 GMT

Now how to join abc1 to abc12?

proc transpose data=abc out=abc12;
by index;
run;

proc print data=abc12;
run;

To get required transformation...

Re: how to loops in following condition

Astounding — Mon, 04 Feb 2019 13:59:21 GMT

Where do you get the color mapping:

0 = g

1 = b

2 = r

That's an important piece of the logic that you have in your head but is not in the data.

Assuming you could extend that list if you needed to, here is a way to automate the process:

data _null_;

call execute ('proc sgplot data=abc;');

do until (done);

set abc1 end=done;

y + 1;

call execute ('series x=index y=Cor ' || put(y, 1.);

call execute('/lineattrs=(color=');

if class=0 then call execute('g');

else if class=1 then call execute('b');

else if class=2 then call execute('r');

call execute ( ');' ) ;

end;

call execute('run;');

stop;

run;

Re: how to loops in following condition

Reeza — Mon, 04 Feb 2019 15:41:52 GMT

@K_coder wrote:

Now how to join abc1 to abc12?
proc transpose data=abc out=abc12;
by index;
run;

proc print data=abc12;
run;
To get required transformation...

Why do you need to join it?

The data attribute map should take care of the colour groupings, which is what you were after, doesn't it?

Re: how to loops in following condition

Astounding — Tue, 05 Feb 2019 05:31:08 GMT

Sorry my fault.

Line 105 needs a closing parenthesis before the final semicolon.

Not working with Cor10, Cor11...

K_coder — Tue, 05 Feb 2019 05:50:38 GMT

Getting error at put when Cor increases

call execute ('series x=index y=Cor ' || put(y, 1.);

After Cor9 and above Error Message-

 + series x=date y=Cor*
                                    _
                                    22
                                    200
 ERROR 22-322: Syntax error, expecting one of the following: ;, /.  
 ERROR 200-322: The symbol is not recognized and will be ignored.

What if column names data(abc) are different

K_coder — Tue, 05 Feb 2019 06:38:23 GMT

consider following condition

data abc;
input index Abc Pqr Xyz A C D E xyz1;
datalines;
1 1 2 4 12 6 10 5 11
2 1 1.5 4 11.5 6.5 10.5 5 13
3 1.5 2 4 12 6 10 5 11
;

data abc1;
input class COORDINATE$;
datalines;
0 Abc
0 Pqr
1 Xyz
2 A
1 C
2 D 
1 E
2 xyz1
;

In above case what should we do?

We have to rename the column names or have another option...

Re: What if column names data(abc) are different

K_coder — Wed, 06 Feb 2019 06:23:50 GMT

Got the solution-

data _NULL_;
call execute('proc sgplot data=abc;');
do until (done);
set abc1;
length s1 $100;
s1 = catx(' ','series x=index y=',COORDINATE);
call execute(s1);
call execute('/lineattrs=(color=');
if class=0 then call execute('g');
else if class=1 then call execute('b');
else if class=2 then call execute('y');
else if class=3 then call execute('r');
call execute ( ');' ) ;
end;
call execute('run;');
run;

Re: What if column names data(abc) are different

Astounding — Wed, 06 Feb 2019 15:00:34 GMT

Excellent!