CATREG

CATREG is available in Sampling and Testing.

CATREG (categorical regression with optimal scaling using alternating least squares) quantifies categorical variables using optimal scaling, resulting in an optimal linear regression equation for the transformed variables. The variables can be given mixed optimal scaling levels, and no distributional assumptions about the variables are made.

CATREG VARIABLES = varlist

/ANALYSIS = 
         depvar [([LEVEL={SPORD**}] [DEGREE={2}] [INKNOT={2}])]
                                            {n  }          {n  }
                         {SPNOM  }  [DEGREE={2}] [INKNOT={2}]
                                            {n  }          {n  }
                         {ORDI   }
                         {NOMI   }
                         {NUME   }
   WITH indvarlist [([LEVEL={SPORD**}] [DEGREE={2}] [INKNOT={2}])]
                                               {n  }          {n  }
                            {SPNOM  }  [DEGREE={2}] [INKNOT={2}]
                                               {n  }          {n  }
                            {ORDI   }
                            {NOMI   }
                            {NUME   }

[/DISCRETIZATION = [varlist [([{GROUPING  }] [{NCAT*={7}}] [DISTR={NORMAL }])]]]
                                                     {n }          {UNIFORM}
                              {EQINTV=n   }
                              {RANKING    }
                              {MULTIPLYING}

[/MISSING = [{varlist}({LISTWISE**})]]
             {ALL**  } {MODEIMPU  }
                                    {EXTRACAT  }

[/SUPPLEMENTARY = OBJECT(objlist)]

[/INITIAL = [{NUMERICAL**}]]
             {RANDOM     }
             {MULTISTART } ({50 }) ('savfile'|'dataset')
                            {n   }
                            {ALL }
             {FIXSIGNS   } (n) ('filename')

[/MAXITER = [{100**}]] 
             {n    }

[/CRITITER = [{.00001**}]]
              {value   }

[/REGULARIZATION = [{NONE**}]] 
                    {RIDGE }   [{(    0,   1.0,  0.02)}] ('filename')
                                {(value, value, value)  }
                    {LASSO }   [{(    0,   1.0,  0.02)}] ('filename')
                                {(value, value, value)  }
                    {ENET  }   [{(    0,   1.0,   0.1)(    0,   1.0,   .02)}] ('filename')
                                {(value, value, value)(value, value, value)}

[/RESAMPLE = [{NONE**   }]] 
              {CROSSVAL }[({10})]
                           {n }
              {BOOTSTRAP}[({50})]
                           {n }

[/PRINT = [R**] [COEFF**] [DESCRIP**[(varlist)]] [HISTORY] [ANOVA**] 
          [CORR] [OCORR] [QUANT[(varlist)]] [REGU] [NONE]]

[/PLOT = [TRANS(varlist)[(h)]] [RESID(varlist)[(h)]] [REGU({valuelist})]]
                                                           {ALL      }

[/SAVE = [TRDATA[({TRA   })]] [PRED[({PRE   })]] [RES[({RES      })]]]
                  {rootname}           {rootname}          {rootname}

[/OUTFILE = [TRDATA('savfile'|'dataset')] [DISCRDATA('savfile'|'dataset')]] .

** Default if the subcommand or keyword is omitted.

This command reads the active dataset and causes execution of any pending commands. See the topic Command Order for more information.

Syntax for the CATREG command can be generated from the Categorical Regression (CATREG) dialog.

Release History

Release 13.0

• The maximum category label length on the PLOT subcommand is increased to 60 (previous value was 20).

Release 17.0

• MULTISTART and FIXSIGNS keywords added to INITIAL subcommand.
• REGULARIZATION subcommand added.
• RESAMPLE subcommand added.
• REGU keyword added to PRINT subcommand.
• REGU keyword added to PLOT subcommand.
• SUPPLEMENTARY categories not occuring in data used to create the model are now interpolated.