topic Re: Large data set in New SAS User

Large data set

superman1 — Sun, 14 Mar 2021 09:54:46 GMT

Re: Recoding Data

Kurt_Bremser — Sun, 14 Mar 2021 12:13:36 GMT

Don't clutter up your data step code with data.

I would create a format that catches all incoming values and translates them to correct ones. Create the format from a dataset in CNTLIN format:

data myformat;
infile datalines dlm=',' dsd truncover end=done;
input (start label) (:$20.);
fmtname = "continent_corr";
type = "C";
output;
if done then do;
  start = "Other";
  label = "************";
  hlo = "O";
  output;
end;
datalines;
North America,North America
South America,South America
Europe,Europe
europe,Europe
New York,North America
;

proc format cntlin=myformat;
run;

Then apply this to your data:

data survey_check;
set survey (keep=continent);
length con_check $20;
con_check = put(continent,continent_corr.);
if con_check = "************";
run;

This dataset will contain all values your format didn't catch yet, so you can include them in the DATALINES.

PS since the OP was rude and disrespectful enough to delete the question, here a rough outline:

A large dataset is received with a variable CONTINENT that should only have "North America", "South America" or "Europe" entered. But, as it goes, lots of other entries are encountered, like "europe" (simple spelling mistake) or "New York" (interesting geography). The method posted above is intended to provide a lasting and easily maintainable remedy.

Re: Large data set

ballardw — Sun, 14 Mar 2021 11:39:24 GMT

@superman1 wrote:

d

Rude to delete the question.

Afraid someone else in your class will get the right answer?

Re: Large data set

superman1 — Sun, 14 Mar 2021 22:31:09 GMT

no I figured it out a different way an didnt want others to spend time to try and figure it out.

Re: Large data set

SASKiwi — Mon, 15 Mar 2021 00:03:28 GMT

@superman1 - If you've found an answer you just mark your post as answered by you then others won't spend any more time on it.

Re: Large data set

mkeintz — Mon, 15 Mar 2021 03:21:42 GMT

@superman1:

As @SASKiwi comments, just post your own solution and mark the problem as solved.

Next time, be @super and satisfy others' curiosity.