topic Re: Creating a new dummy variable for missing values in New SAS User

Creating a new dummy variable for missing values

brookedanel — Mon, 12 Nov 2018 17:49:25 GMT

I'm trying to do a secondary analysis on a very large dataset that has a lot of missing values. I'd like to see if there is any pattern within the missing values. I'd like to create a new dummy variable for missing so 0=missing and 1=not missing. I'm not sure how to go about this. I've created a smaller dataset that narrow down my predictor and outcome variables of interest. All of the variables are numeric/continuous and each variable has missing values. Any suggestions are greatly appreciated!

Re: Creating a new dummy variable for missing values

novinosrin — Mon, 12 Nov 2018 17:55:44 GMT

do you want as simple as this?

data have;
do i=1 to 100;
num=int(ranuni(76567)*100);
if mod(i,5)=0 then call missing(num);
output;
end;
run;

data want;
set have;
dummy_num=num ne .;
run;

Re: Creating a new dummy variable for missing values

brookedanel — Mon, 12 Nov 2018 17:58:23 GMT

Thank you so much! Can you just explain what this step does

num=int(ranuni(76567)*100);
if mod(i,5)=0 then call missing(num);

Re: Creating a new dummy variable for missing values

novinosrin — Mon, 12 Nov 2018 18:03:56 GMT

ignore the data have step. I wanted to create some sample data with random numbers as you didn't provide a sample

data want is what you need

For your interest,

Using ranuni function I created 100 random numbers between 0 and 100. For every 5th or in other words for every multiple of 5th value, i assigned them as missing. This is what data have does,

Data want is a simple boolean expression i.e if num is missing then true else false in the form of 1s and 0s

Re: Creating a new dummy variable for missing values

FreelanceReinh — Mon, 12 Nov 2018 18:34:26 GMT

Hi @brookedanel,

If all you need are numbers and percentages of missing values for each variable or similar results for combinations of variables, you don't need dummy variables. Just use a format.

Example:

proc format;
value miss
._-.z = 'missing'
other = 'non-missing';

value $miss
' '   = 'missing'
other = 'non-missing';
run;

proc freq data=yourdata;
format _numeric_ miss. _character_ $miss.;
tables _all_ / missing; /* or replace _all_ with another variable list */
run;

Edit: The output (here for SASHELP.HEART) will look something like this:

The FREQ Procedure

                                        Cumulative    Cumulative
Status         Frequency     Percent     Frequency      Percent
----------------------------------------------------------------
non-missing        5209      100.00          5209       100.00


                         Cause of Death

                                        Cumulative    Cumulative
DeathCause     Frequency     Percent     Frequency      Percent
----------------------------------------------------------------
missing            3218       61.78          3218        61.78
non-missing        1991       38.22          5209       100.00


                        Age CHD Diagnosed

                                        Cumulative    Cumulative
 AgeCHDdiag    Frequency     Percent     Frequency      Percent
----------------------------------------------------------------
missing            3760       72.18          3760        72.18
non-missing        1449       27.82          5209       100.00

... portions of the output are not shown ...

Edit 2: Here is an example using a combination of variables:

proc freq data=sashelp.heart;
format _numeric_ miss. _character_ $miss.;
tables status*deathcause*agechddiag / missing list;
run;

Output:

                                                                      Cumulative    Cumulative
Status         DeathCause      AgeCHDdiag    Frequency     Percent     Frequency      Percent
----------------------------------------------------------------------------------------------
non-missing    missing        missing            2663       51.12          2663        51.12
non-missing    missing        non-missing         555       10.65          3218        61.78
non-missing    non-missing    missing            1097       21.06          4315        82.84
non-missing    non-missing    non-missing         894       17.16          5209       100.00

Re: Creating a new dummy variable for missing values

brookedanel — Mon, 12 Nov 2018 18:25:11 GMT

Thank you so much @FreelanceReinh. I'm going to give this a shot. It's similar to the code I was using but I see now where I missed something. I appreciate the quick reply.