Proc Logistics Score Statement : fifstat option explain Misclassification (error) rate

[AmitKB]

Hi all,

            can anyone explain the Misclassification (error) rate statistics we get when we use fitstat option in score statement in proc logistics. I looked up the documentation but could not understand it.

            I appreciate all your help.

Regards,

Amit

  Proc logistics data=train;

     model target (event = '1') = var1 var2 var3;

    score data= validate out=Scored fitstat;

run;

[Season]

Hello, there. It has been many years since you raised the question. I wonder if that problem has been solved right now. I am also building a logistic regression model and has been working on the issue of misclassification error for some time. As SAS Help shows, the formula of misclassification rate is . Simply speaking, according to this formula, the proportion of observations that were misclassified is designated as the misclassification rate. In SAS Help, it is stated that an observation is classified into the level with the largest probability.

For instance, suppose the dependent variable has two levels, 0 and 1. If P(Y=1) (posterior probability) of an observation is not smaller than P(Y=0) (posterior probability), then SAS deems that according to the model, the event coded by Y=1 happens to the observation; and vice versa.