Categorize connected nonmissing values

[mgrasmussen]

Dear SAS expert

I would like to categorize "connected" 1's according to have_var (a 1 increment for every new connected 1s). I have tried to get it to work but no succes thus far. I have managed to create a count which starts over with each category (current_succes) but this does not suffice. Can anyone help me with the code for want_var given the following example data?

data have;
input have_var current_succes want_var;
datalines;
. . .
. . .
1 1 1
1 2 1
. . .
1 1 2
. . .
1 1 3
1 2 3
1 3 3
. . .
;
run;

Thank you

[PaigeMiller]

How can we determine what you are doing when there are lines such as 

...

in your input data? Can you give us actual (or made-up) data in place of the ... ? Or can you explain more clearly what you are looking for?

[PeterClemmensen]

I don't quite understand this. 

Do you want want_var to increment when have_var = 1 and the previous have_var = 1 ? 

Otherwise, please explain further.

[mgrasmussen]

Every time there is a 1 or consecutive 1s I want to introduce a new number. The dataset is sorted. Actually what I really want is just to categorize 1s (only 1 or consecutive) in unique categories. It could in principle also be using letters. For example:

have_var number_categorization letter_categorization

. . .

1 1 A

. . .

1 2 B

1 2 B

1 2 B

. . .

. . .

. . .

1 3 C

1 3 C

The 2 variables to the right above displays the data that I would like to have - I don't have these two variables. The data that I have is only the variable on the left.

Missing values could be changed to 0's if this in terms of code would make a difference.

Am I still failing to make sense?

[PeterClemmensen]

We're getting there 😉

So, in between the 1's, do you want missing values or does that not matter?

[mgrasmussen]

I apologize for being unclear.

Yes, missing values between the 1's 🙂

The code needs to work for data exactly like the example I provided.

[PeterClemmensen]

Have you seen my code below?

[PeterClemmensen · Accepted solution]

If so then how about

data have;
input v;
datalines; 
2
1
3
1
1
1
4
5
6
1
1
;

data want(drop = ww);
   set have;
   by v notsorted;
   if v ne 1 then w = .;
   if v = 1 and first.v then do;
      w = ww + 1;
	  ww + 1;
   end;
   retain w;
run;

Result:

Obs v  w 
1   2  . 
2   1  1 
3   3  . 
4   1  2 
5   1  2 
6   1  2 
7   4  . 
8   5  . 
9   6  . 
10  1  3 
11  1  3 

[mgrasmussen]

Yes, thanks. However, for some reason my SAS is not working right now (poor timing).

But would your code still work given the following data:

data have;

input v;

datalines;

.

1

.

1

1

1

.

.

.

1

1

;

run;

Notice that there has to be . (missings) between the 1s in the input data.

[PaigeMiller]

---

@mgrasmussen wrote:

Every time there is a 1 or consecutive 1s I want to introduce a new number. The dataset is sorted. Actually what I really want is just to categorize 1s (only 1 or consecutive) in unique categories. It could in principle also be using letters. For example:

have_var number_categorization letter_categorization

. . .

1 1 A

. . .

1 2 B

1 2 B

1 2 B

. . .

. . .

. . .

1 3 C

1 3 C

The 2 variables to the right above displays the data that I would like to have - I don't have these two variables. The data that I have is only the variable on the left.

Missing values could be changed to 0's if this in terms of code would make a difference.

Am I still failing to make sense?

---

You are failing to make sense to me. What are the rules? Please provide example data without the ... (triple dots), so we can see exactly what this would look like, data and desired output.