The DiscGen06.lng Model

Discriminant analysis by linear programming

View the model

Download the model

Discriminant analysis by linear programming;

There are two categories of observations, type 0 and type 1,

which we wish to distinguish based on explanatory variables;

Key idea: Find a scoring formula:

score( i) = beta0 + beta1*xdat( k, 1) + beta2*xdat( i, 2) + ...

and try to enforce:

score( i) < 0 if observation i is of type 0,

score( i) > 0 if observation i is of type 1;

Keywords:

Discriminant Analysis | Binary Choice | Categorical variables | Random forest |