comparing rows in a dataset

[Anita_n]

Dear all,

I have a dataset that looks like this

pid tum_typ sex   date_of_diagnosis

1     2           2      10/04/2015

1     2           2       01/06/2010

1     2           2      11/03/2019

2     2          1        09/09/2000

2     2          1        05/08/2018

3     2          1        02/10/2020

4     2          2        11/02/2001

4     2          2        15/12/1998

5     2         1

5     2         1          07/07/2020

5     2         1          06/01/2020

I wish to create a new variable called order_of_tum which indicates the 1., 2. 3., etc of tumor a patient has grouped by id and date. The result should look like this :

That means I compare the first and second row and if date 1 is less than date2 then date1 =1. and date2=2.

pid tum_typ sex   date_of_diagnosis  order_of_tum

1     2           2      01/06/2010                1

1     2           2      10/04/2015                2

1     2           2      11/03/2019                3

2     2          1        09/09/2000              1

2     2          1        05/08/2018              2

3     2          1        02/10/2020              1

4     2          2        15/12/1998              1

4     2          2        11/02/2001               2

5     2         1          06/01/2020             1

5     2         1          07/07/2020             2

5     2         1                                         3

Thanks

[PeterClemmensen · Accepted solution]

I think this is what you want

data have;
infile datalines missover;
input pid tum_typ sex date_of_diagnosis : ddmmyy10.;
format date_of_diagnosis ddmmyy10.;
datalines;
1 2 2 10/04/2015
1 2 2 01/06/2010
1 2 2 11/03/2019
2 2 1 09/09/2000
2 2 1 05/08/2018
3 2 1 02/10/2020
4 2 2 11/02/2001
4 2 2 15/12/1998
5 2 1           
5 2 1 07/07/2020
5 2 1 06/01/2020
;

proc sort data = have;
   by pid date_of_diagnosis;
run;

data want;
   set have;
   by pid date_of_diagnosis;
   if first.pid then order_of_tum = 0;
   if first.date_of_diagnosis then order_of_tum + 1;
run;

[Anita_n]

Thank you

[Anita_n]

Hi what happens when I have the same dates within a group? In that case don't want it to count further but out the number as previous.

for example if patient 1,  there is a 4th row  with 

1 2 2 11/03/2019

I don't want  to have order_of_tum like (1, 2, 3, 4)  but  (1, 2 ,3 ,3) any solution to that?