How to use IF THEN function to count on each observation separately

[Noleen]

I need to count how many numbers starting with '1' '2' '3's in each observation.

There are 700 variables and 50000 observations.

I use the following code,

data want;

set have;

array chars $ chars1-char700;

array counts counts1-counts3;

do over chars ;

if substr(chars,1,1)=1 then counts1+1;

if substr(chars,1,1)=2 then counts2+1;

if substr(chars,1,1)=3 then counts3+1;

end;

run;

but the counts on each observation is adding on top of the result (counts) from the last observation.

How can I count them for each observation? (the counts need to star from 0 for each observation)

Many thanks

[Reeza]

Do you only have 1,2,3?

To answer your initial question counts1+1 creates an implicit retain, change the counter to counts1=counts1+1 or explicitly set the variables to 0 at the beginning of each loop

[Noleen]

Thanks Reeza,

I tried count1=count1+1, it doesn't count the number at all. I don't know why. Do I need to set count1=0 at the beginning?

BTW, how can I set counts back to 0 at the beginning of the loop?

Cheers

[Haikuo]

On top of Reeza's suggestions, you can also set counts missing for every data step iteration:

call missing (of counts(*)); Some simplification has also been suggested in the following code:

data want;

     set have;

     array chars chars1-chars700;

     array counts counts1-counts3;

     call missing(of counts(*));

     do over chars;

           counts(input(first(chars),1.))+1;

     end;

run;

[Noleen]

Thanks Hai,

I'm not sure I understand it.

Is the count missing can reset variables to 0 ?

Cheers,

[Reeza · Accepted solution]

data want;

set have;

array chars $ chars1-char700;

array counts counts1-counts3;

counts1=0;counts2=0;counts3=0;

do over chars ;

if substr(chars,1,1)=1 then counts1+1;

if substr(chars,1,1)=2 then counts2+1;

if substr(chars,1,1)=3 then counts3+1;

end;

run;

[Noleen]

Hi Reeza;

It still sum up the counts..

[Noleen]

Hi Reeza,

I miss typed the name.

It works!! Thanks a lot!