436952_en-US

liliconcom

User Input List for Regression Report with Interactions

I have an input box where users can select columns to be run in a regression report. This part works great. I then insert the list of columns into a list. I would like to take the list of columns and make different interactions that can be inserted into a regression report. Specifically, I am interested in slightly more complicated interactions (i.e. Column 1 * Column 2 ^2) When I try to put anything but just the generic list in the regression report, it doesn’t run the report. What do I do to create these interactions in the report?

 

The user input list readout looks like this: {"Column 1", "Column 2", "Column 3"}

 

 

platform = dt << Fit Model(

 

       Y( :Char(variable)),

 

             Effects(

 

                           What should go here?????

 

       ),

 

     

 

       Personality( "Generalized Regression" ),

 

       Generalized Distribution( "Normal" ),

 

       Run(

 

             Fit( Estimation Method( Two Stage Forward Selection ), Validation Method( AICc ), Enforce Heredity )

 

       )

 

);

 

 

DOEScriptingRe: User Input List for Regression Report with Interactions

Here is one way to handle this

names default to here(1);
dt=data table("big class");
variable = "Age";
// If you build your list as expressions the code below works
effectsList = {height, weight, weight * height};

Fit Model(
 Y( :char(variable) ),
 Effects( eval(effectsList) ),
 Personality( "Standard Least Squares" ),
 Emphasis( "Effect Leverage" ),
 Run(
  :char(variable) << {Summary of Fit( 1 ), Analysis of Variance( 1 ),
  Parameter Estimates( 1 ), Scaled Estimates( 0 ),
  Plot Actual by Predicted( 1 ), Plot Regression( 0 ),
  Plot Residual by Predicted( 1 ), Plot Studentized Residuals( 0 ),
  Plot Effect Leverage( 1 ), Plot Residual by Normal Quantiles( 0 ),
  Box Cox Y Transformation( 0 )}
 )
);

/*
// If you build your list using strings
effectsList = {"height", "weight", "weight * height"};
// You can convert the elements back to expressions by
// using the Parse() function
For( i = 1, i <= N Items( effectsList ), i++,
 effectsList[i] = Parse( effectsList[i] );
);

Re: User Input List for Regression Report with Interactions

The Scripting Index has an example for Effects - copied in below. 

 

Names Default To Here( 1 );
dt = Open( "$SAMPLE_DATA/Dogs.jmp" );
obj = Fit Model(
 Y( :LogHist0, :LogHist1, :LogHist3, :LogHist5 ),
 Effects( :drug, :dep1, :drug * :dep1 ),
 Personality( Manova ),
 Run( Response Function( Sum ) )
);

 

Re: User Input List for Regression Report with Interactions

Here is another way using expressions. There are more solutions based on expressions than just this example.

 

Names Default To Here( 1 );

dt = Open( "$SAMPLE_DATA/Dogs.jmp" );

predictor = { :drug, :dep1 };
n = N Items( predictor );

effectExpr = Expr( Effects() );
For( x1 = 1, x1 < n, x1++,
 Insert Into( effectExpr, predictor[x1] );
 For( x2 = 2, x2 <= n, x2++,
  Eval(
   Substitute(
    Expr( Insert Into( effectExpr, Expr( xxx1 * xxx2 ) ) ),
    Expr( xxx1 ), predictor[x1],
    Expr( xxx2 ), predictor[x2]
   )
  )
 );
);
Show( Name Expr( effectExpr) );

Eval(
 Substitute(
  Expr(
   obj = dt << Fit Model(
    Y( :LogHist0, :LogHist1, :LogHist3, :LogHist5 ),
    eee,
    Personality( Manova ),
    Run( Response Function( Sum ) )
   );
  ),
  Expr( eee ), Name Expr( effectExpr )
 )
);