Programming the statistical procedures from SAS

Variable selection in Proc Logistic - dealing with missing values

Reply
Contributor
Posts: 73

Variable selection in Proc Logistic - dealing with missing values

[ Edited ]

Dear All,

 

Here I have a question about variale selection. Say I have the following data set:

 

Age    Sex    weight     Pass

15          0           70           1

17          1              -           0

16           -           60           1   ;

 

And I want to use logistic regression to predice pass using age, sex and weight . also I want to use backwards variable selection to selsect significant covariate. So I coded as following:

 

proc logistic data=have descending;

model pass= age sex weight /selection=backward fast;

run;

 

However, since I have missing data here. The program only use the entry that has complete observation based on full model, therefore the first enrty.

 

But I want it to enclude more observation as eleminate covariates, that is when model only with age and sex, i want it to use first two entries. Is there a way to realize this in SAS?

 

Thank you very much!!!

 

Best,

 

 

 

HereH

Grand Advisor
Posts: 10,062

Re: Variable selection in Proc Logistic - dealing with missing values

The procedure will only use records that have values for all the model variables.

 

You might try imputation to replace missing values. There are a number of ways but if you have significant percentage of missing values for a given variable then your resulting model is going to be suspect.

 

What percentage of your records are missing one or more of the variables?

Respected Advisor
Posts: 4,606

Re: Variable selection in Proc Logistic - dealing with missing values

You could screen your predictors with PROC HPSPLIT. It offers three methods for dealing with missing predictors. The output decision tree is a crude model but will show you which variables are the most important predictors.

PG
Ask a Question
Discussion stats
  • 2 replies
  • 143 views
  • 2 likes
  • 3 in conversation