Randomly select records based on sum of a column

[Hima]

I need to create a new data set using the Have dataset. The conditons are as follow:

1. Include all observations with Column C = C

2. Select random observations with Column C = T where sum of  TTL_PLST less than or equal to 20. ** Note: The records highlighted in Red make a sum of 20 where Column C = T.**

Dataset Have

Column A	Column B	Column C	Column D	TTL_PLST	N
123456789	1A	C	NP	1	1
123456789	1A	C	NP	2	33
123456789	1A	C	NP	3	21
123456789	1A	C	NP	4	2
123456789	1A	C	NP	5	64
123456789	1A	C	NP	6	2
123456789	1A	C	NP	7	97
123456789	1A	T	NP	8	10
123456789	1A	T	NP	9	1
123456789	1A	T	NP	10	3
123456789	1A	T	NP	11	1
123456789	1A	T	NP	3	1
123456789	1A	T	NP	13	2
123456789	1A	T	NP	14	264
123456789	1A	T	NP	15	1

Want Dataset:

Column A	Column B	Column C	Column D	TTL_PLST	N
123456789	1A	C	NP	1	1
123456789	1A	C	NP	2	33
123456789	1A	C	NP	3	21
123456789	1A	C	NP	4	2
123456789	1A	C	NP	5	64
123456789	1A	C	NP	6	2
123456789	1A	C	NP	7	97
123456789	1A	T	NP	8	10
123456789	1A	T	NP	9	1
123456789	1A	T	NP	3	1

[Linlin · Accepted solution]

how about:

data have ;

input Column_A $9. Column_B $ Column_C $ Column_D $ TTL_PLST N ;

cards;

123456789 1A C  NP 1 1

123456789 1A C  NP 2 33

123456789 1A C  NP 3 21

123456789 1A C  NP 4 2

123456789 1A C  NP 5 64

123456789 1A C  NP 6 2

123456789 1A C  NP 7 97

123456789 1A T  NP 8 10

123456789 1A T  NP 9 1

123456789 1A T  NP 10 3

123456789 1A T  NP 11 1

123456789 1A T  NP 3 1

123456789 1A T  NP 13 2

123456789 1A T  NP 14 264

123456789 1A T  NP 15 1

;

proc sort data=have out=have1(where=(column_c='T'));

  by TTL_PLST;

run;

data have1(drop=total);

  set have1;

  retain total;

   total+TTL_PLST;

  if total <=20;

run;

data want;

   set have (where=(column_c='C'))

     have1;

run;

Linlin

[Hima]

Thanks for your quick reply Linlin. Looking at the original data, I totally lost what I was doing. So, I had to post it here.

[Ksharp]

Can't understand what you mean about 'Randomly Select observation'