Hello,
data _null_;
x='A,B,C,D,E,F';
do n=4, 5;
position=length(prxchange(cats('s/^(([^,]*,){',n,'}).*$/$1/'),1,x));
put n= position=;
end;
run;
... View more
What does your input look like, or an example, and what should the output look like?
There can be more than one interpretation of your subject line.
It could be "is the record order value even/odd"
It could be the record number is even/odd (not always the same thing with deleted information)
Is the value of some variable, or combination of variables even/odd.
Is the explicit value of some variable (or variables) the text "even" or "odd".
And I suspect that if I spent some time I might come up with a few more. So something specific might be a good idea.
... View more
Well, if have has 600 million records but only 1 million unique key value, you would would want to avoid sorting the 600 million records.
If that is the scenario, you could read WANT sequentially only once, use a datastep with hashing to look up whether you have a match.
If so, insert the first match into the hash table (of course the other non-key saved variables might not be want you want).
At then at the end you output the hash table to a physical table.
You need to ensure that the million full rows will fit into memory. If that is not possible then
create a macro that read HAVE in n pieces using options firstobs and obs.
Then you need a final part that re-reads the reduced pieces and consolidates them into a final HAVE.
... View more
Just for fun.
Data have;
input id;
cards;
-1
-2
-3
10
20
30
;
proc sql;
select (select sum(id) from have where id>0) as positive,
(select sum(id) from have where id<0) as negative
from have(obs=1);
quit;
... View more
Well, if have has 600 million records but only 1 million unique key value, you would would want to avoid sorting the 600 million records.
If that is the scenario, you could read WANT sequentially only once, use a datastep with hashing to look up whether you have a match.
If so, insert the first match into the hash table (of course the other non-key saved variables might not be want you want).
At then at the end you output the hash table to a physical table.
You need to ensure that the million full rows will fit into memory. If that is not possible then
create a macro that read HAVE in n pieces using options firstobs and obs.
Then you need a final part that re-reads the reduced pieces and consolidates them into a final HAVE.
... View more
Cid date
101 01jan2019
102 02jan2019
103 02jan2019
104 01feb2019
105 02feb2019
101 05feb2019
106 07feb2019
108 01mar2019
104 05mar2019
107 01apr2019
109 02may2019
106 03may2019
102 04may2019
109 05jun2019
.
.
.
the above data we have
i want customer who visited twice in any consecutive 3months.It means customer who visited 2times in (jan feb mar) .means in three months customer need to visit atleast 2times.
for the above who visited twice in three months is
cid
101
104
109
but in data we have many observations....
for example i have given few records to underdstand senario..
... View more
Typing in all caps is yelling on the internet and it's actually hard to read. Please format your questions.
To answer your question use the LIBNAME() function.https://documentation.sas.com/?docsetId=lefunctionsref&docsetTarget=p1bq8nyxm7y1ygn1i4vyf82z68ls.htm&docsetVersion=9.4&locale=en
RC is the return code which indicates if the function was successful.
DATA A;
INPUT LIBREF$ PATH$;
rc = libname(libref, path);
CARDS;
MYLIB C:\SAS\SAI
MYSAS D:\SAS\RAJA
MYXYZ E:\SAS\RANI
;
RUN;
... View more
Hi @Saikiran_Mamidi
You can use the PRXMATCH function for example:
data want;
set have;
if prxmatch('/^\d{5}[a-z]$/',strip(names)) then output;
run;
It looks for the following pattern: ^\d{5}[a-z]$
^ = beginning of the string
\d{5} = 5 digits
[a-z] = one letter (case insensitive -> cf "i" modifier)
$ = end of the string
strip() enables to take into account trailing blanks.
You could also use:
if prxmatch('/^\d{5}[a-z]\s*$/i',names) then output;
-> \s means a blank and * means 0,1 or more times -> looks for trailing blanks
... View more
Hi @Saikiran_Mamidi
In addition to PROC SQL, you can also easily do that using a PROC SORT with the NODUPKEY option.
You can output:
- a dataset with no duplicate records (out = )
- a dataset with duplicate records (dupout = )
proc sort data=have out=no_dupkey dupout=dupkey nodupkey;
by _all_;
run;
Dataset 'no_dupkey' Dataset 'dupkey'
... View more
You can see that there are simple versions and there are complex (but more flexible) versions. If you use a macro-based approach, here are a couple of changes to simplify and speed up the program.
The macro contains:
%do I=1 %to #
if %eval(&n*(&i-1)) <_n_ <=%eval(&n*&I) then output SAMPLE_&I;
%end;
This would work equally well:
if _n_ <= &n then output SAMPLE_1;
%do I=2 %to #
else if _n_ <= %eval(&n*&I) then output SAMPLE_&I;
%end;
... View more
That's not so hard:
data want;
set have;
length _subt $2;
do _N_=1 to countw(subt);
_subt=scan(subt,_N_,',');
output;
end;
drop subt;
rename _subt=subt;
run;
... View more
Hi @Saikiran_Mamidi
you can also use a PROC FORMAT with the PICTURE statement:
data have;
input A B;
datalines;
123.22 44.21
;
run;
proc format;
picture myformat low-high = 9999999.99;
run;
data want;
set have;
format A B myformat.;
run;
... View more
