/* Note that the MDC procedure does not include the intercept term
automatically like other regression procedures. The dependent variable
decision takes the value 1 when a specific alternative is chosen;
otherwise it takes the value 0. Each individual is allowed to choose one
and only one of the possible alternatives. In other words, the variable
decision takes the value 1 one time only for each individual.
If each individual has three elements (1, 2, and 3) in the choice set,
the NCHOICE=3 option can be specified instead of CHOICE=(mode 1 2 3).
Consider the following trinomial data from Daganzo (1979). The original
data (origdata) contains travel time (ttime1-ttime3) and choice (choice)
variables. ttime1-ttime3 are the travel times for three different modes
of transportation, and choice indicates which one of the three modes is
chosen. The choice variable must have integer values.  This program is 
taken directly from the Proc MDC help file.  */ 

data origdata; 
      input ttime1 ttime2 ttime3 choice @@; 
      datalines; 
   16.481  16.196  23.89   2  15.123  11.373  14.182  2 
   19.469  8.822   20.819  2  18.847  15.649  21.28   2 
   12.578  10.671  18.335  2  11.513  20.582  27.838  1  
   10.651  15.537  17.418  1  8.359   15.675  21.05   1  
   11.679  12.668  23.104  1  23.237  10.356  21.346  2  
   13.236  16.019  10.087  3  20.052  16.861  14.168  3  
   18.917  14.764  21.564  2  18.2    6.868   19.095  2  
   10.777  16.554  15.938  1  20.003  6.377   9.314   2  
   19.768  8.523   18.96   2  8.151   13.845  17.643  2  
   22.173  18.045  15.535  1  13.134  11.067  19.108  2  
   14.051  14.247  15.764  1  14.685  10.811  12.361  3  
   11.666  10.758  16.445  1  17.211  15.201  17.059  3  
   13.93   16.227  22.024  1  15.237  14.345  19.984  2 
   10.84   11.071  10.188  1  16.841  11.224  13.417  2 
   13.913  16.991  26.618  3  13.089  9.822   19.162  2 
   16.626  10.725  15.285  3  13.477  15.509  24.421  2 
   20.851  14.557  19.8    2  11.365  12.673  22.212  2 
   13.296  10.076  17.81   2  15.417  14.103  21.05   1 
   15.938  11.18   19.851  2  19.034  14.125  19.764  2 
   10.466  12.841  18.54   1  15.799  16.979  13.074  3 
   12.713  15.105  13.629  2  16.908  10.958  19.713  2 
   17.098  6.853   14.502  2  18.608  14.286  18.301  2 
   11.059  10.812  20.121  1  15.641  10.754  24.669  2 
   7.822   18.949  16.904  1  12.824  5.697   19.183  2 
   11.852  12.147  15.672  2  15.557  8.307   22.286  2   
   ;

   /*  Here we convert the data to satisify the data format
       of Proc MDC.  */

data newdata(keep=pid decision mode ttime); 
      set origdata; 
      array tvec{3} ttime1 - ttime3; 
      retain pid 0; 
      pid + 1; 
      do i = 1 to 3; 
         mode = i; 
         ttime = tvec{i}; 
         decision = ( choice = i ); 
         output; 
      end; 
   run;

   proc print data = newdata;

   run;

/*  Now we run the multinomial logit on the Daganzo data.  */
 
proc mdc data=newdata; 
      model decision = ttime / type=clogit nchoice=3 
         optmethod=qn covest=hess; 
      id pid; 
   run;

/*  Here we set up the out-of-sample data for predicting
    the choice of the 51st individual.  */

data extra; 
      input pid mode decision ttime; 
      datalines; 
   51  1  .   5.0 
   51  2  .  15.0 
   51  3  .  14.0 
   ;


   data extdata; 
      set newdata extra; 
   run; 

/* Now we use the estimated multinomial logit model to predict the out-of-sample
   data.  */
 
   proc mdc data=extdata; 
      model decision = ttime / 
            type=clogit covest=hess  
            nchoice=3; 
      id pid; 
      output out=probdata pred=p;       
   run; 

/*  Now we print out the prediction for the 50th and 51st individuals,
    the 51st individual being the out-of-sample individual.  */
 
   proc print data=probdata( where=( pid >= 49 ) ); 
      var mode decision p ttime;  
      id pid; 
   run;