Replicating the function of Optimal Binning Function (SAS EM)

[willfagerland]

Hello,

I have a question about how to interpret the SAS EM output on Analysis of Maximum Likelihood Estimates from linear regression model. 

Our model has 2 predictors (i.e., var1 and var2). We used the opt binning transformation on var2. We used 4 bins for VAR2.  The coefficients from my selected regression model for variables are:

INTERCEPT = 1.234

VAR1 = 0.2

OPT_VAR2 01 =  0.111

OPT _VAR2 02 = 0.22

OPT _VAR2 03 = 0.32 

OPT_VAR2 04 BIN was not included in the selected model. 

We are hoping for help to determine the missing piece of the following equation:

Outcome = intercept + 0.2VAR1 + ??? 

Thanks,

Will

[Reeza]

Getting the scoring code is usually an easier way, but it's also just;

outcome = intercept + 0.2 * VAR1 + 0.111*opt_var01 + 022*opt_var02 + 0.32*opt_varo03;

Your 04 is included as the reference level and in the intercept. (not included in the select model because you need n-1 dummy variables for 4 levels). So you'll need to bin the data to how you did before, but again, the scoring code does that for you.