topic Re: How to set datasets with same names but different labels in SAS Programming

How to set datasets with same names but different labels

Roger5 — Wed, 26 Feb 2020 16:33:27 GMT

Hi,

I am trying to set datasets with same names that have different labels. The names are not what I am interested in. I want to keep the labels. Do you have any idea how I can do this?

e.g.

Dataset1

name = city1, label=city_name_very_too_long_to_be_a_nameBoston

name = city2, label=city_name_very_too_long_to_be_a_nameNewYork

Dataset2

name = city1, label=city_name_very_too_long_to_be_a_nameChicago

name = city2, label=city_name_very_too_long_to_be_a_nameCapeCod

Thanks!

Re: How to set datasets with same names but different labels

PaigeMiller — Wed, 26 Feb 2020 16:38:00 GMT

It's not clear to me what "I want to keep the labels" means in this application. The resulting data set can have only one label per variable, so what is the desired output?

Re: How to set datasets with same names but different labels

Tom — Wed, 26 Feb 2020 16:45:08 GMT

When you combine datasets that contain the same variable then the label for the variable is set by the first dataset that actually has a label attached. You can override it by adding a label statement to the step that combines the datasets.

data one ;
  call missing(x,y,z);
  label x='X in ONE' y='Y in ONE';
run;
data two ;
  call missing(x,y,z);
  label Y='Y in TWO' z='Z in TWO';
run;
data both;
 set one two;
run;
proc contents data=both; run;

So if your preferred labels are on DATASET2 then list if first.

Re: How to set datasets with same names but different labels

Roger5 — Wed, 26 Feb 2020 16:46:50 GMT

Hi,

Dataset1

name = city1, label=city_name_very_too_long_to_be_a_nameBoston, n=20

name = city2, label=city_name_very_too_long_to_be_a_nameNewYork, n=10

Dataset2

name = city1, label=city_name_very_too_long_to_be_a_nameChicago, n=5

name = city2, label=city_name_very_too_long_to_be_a_nameCapeCod, n=0

If I do:

data dataset3;

set dataset1 dataset2;

run;

Dataset3;

City1 (label=city_name_very_too_long_to_be_a_nameChicago) City2 (label=city_name_very_too_long_to_be_a_nameCapeCod)

20 10

5 0

I am loosing the information related to my first Dataset1 (Boston and NewYork). Therefore, in Dataset3, I am assuming that 20 patients are coming from Chicago while they are coming from Boston.

Hope this is more clear.

Re: How to set datasets with same names but different labels

Tom — Wed, 26 Feb 2020 16:49:30 GMT

@Roger5 wrote:

Hi,

Dataset1

name = city1, label=city_name_very_too_long_to_be_a_nameBoston, n=20

name = city2, label=city_name_very_too_long_to_be_a_nameNewYork, n=10

Dataset2

name = city1, label=city_name_very_too_long_to_be_a_nameChicago, n=5

name = city2, label=city_name_very_too_long_to_be_a_nameCapeCod, n=0

If I do:

data dataset3;

set dataset1 dataset2;

run;

Dataset3;

City1 (label=city_name_very_too_long_to_be_a_nameChicago) City2 (label=city_name_very_too_long_to_be_a_nameCapeCod)

20 10

5 0

I am loosing the information related to my first Dataset1 (Boston and NewYork). Therefore, in Dataset3, I am assuming that 20 patients are coming from Chicago while they are coming from Boston.

Hope this is more clear.

Sounds like you want to rename the variables when the labels are different.

Might be best to first transpose the data before combining it so that the data (name of city) is in a VARIABLE and not in metadata (variable LABEL).

Re: How to set datasets with same names but different labels

PaigeMiller — Wed, 26 Feb 2020 16:50:14 GMT

I still don't think you say what you WANT to happen, you just explain what does happen to the labels is not acceptable.

Perhaps you want to create a variable where it takes on values such as Boston, Chicago, New York, etc. Would that work for you?