217471_en-US

liliconcom

Re: Adding augmented data points to the end of an already existing data table through JSL

1. You have a reference in your JSL to a dt2 data table.

   PredictedResidual = dt2 << Neural(

   I assume that is supposed to be "dt"
2. The Add Rows() function expects a numeric value specified, which indicates the number of rows to add.  Your reference has no value, so it isn't going to add any rows

   dt << add rows();

Adding augmented data points to the end of an already existing data table through JSL

I currently have a script that generates a random column of X's, calculates a column of Y's based on the X's, makes a predicted Y column, a residual column (Predicted Y - Y), and a predicted Y column. From there, my code selects the rows with residuals lower than 3 and deletes them.

 

I want to then add more rows of augmented data points, and do the cycle again with the intent of making the residuals lower. Below is my code:

 

 

// 1. Creates X1 column, and corresponding Y1 and Ystat columns
Close All( Data Tables, NoSave );
dt = New Table( "SampleRun 3",
 Add Rows( 100 ),
 New Script(
  "Ystat vs. X1",
  Graph Builder(
   Variables( X( :X1 ), Y( :Ystat ) ),
   Elements( Points( X, Y, Legend( 5 ) ) )
  )
 ),
 New Column( "X1", 

  Formula( Random Integer( 50 ) )
 ),
 New Column( "Y1", 

  Formula( If( :X1 < 25, 2 * :X1, 25 <= :X1 <= 30, 3 * :X1, 4 * :X1 ) )
 ),
 New Column( "Ystat", 

  Formula( :Y1 )
 )

);
// 2. Creates predicted Ystat column, saves formulas into same data table (first iteration)
PredictedYstat = dt << Neural(
 Y( :Ystat ),
 X( :X1 ),
 Informative Missing( 0 ),
 Validation Method( "Holdback", 0.3333 ),
 Fit( NTanH( 3 ) )
);
PredictedYStat << Save Profile Formulas;

// 3. Adds residual column
dt << New Column( "Residual 1", Formula( Abs( :Predicted Ystat - :Ystat ) ) );
Column( dt, "X1" ) << Suppress Formula Eval();
Column( dt, "Y1" ) << Suppress Formula Eval();
Column( dt, "Ystat" ) << Suppress Formula Eval();
dt << Select where( :Residual 1 <= 3 );
dt << delete rows;
// 4. Creates and predicts residual column
PredictedResidual = dt2 << Neural(
 Y( :Residual 1 ),
 X( :X1 ),
 Informative Missing( 0 ),
 Validation Method( "Holdback", 0.3333 ),
 Fit( NTanH( 3 ) )
);
PredictedResidual << Save Profile Formulas;
// 5. Adds new data points, cycle starts over from 2 (lets user choose how many)
// Iteration 2
DOE( Augment Design, X( :X1 ), Y( :Ystat ), {Make Design, Make Table} );
dt << add rows();

 

 

But the last chunk of code (where it is supposed to add rows) doesn't do anything. Any suggestions?

 

Thanks,

Kyle

 

Re: Adding augmented data points to the end of an already existing data table through JSL

Yes, it was supposed to be dt instead of dt2. Thank you for the correction. If add rows(); expects a specified value, how would I go about adding more X values from an augmented design to the end of the data table "Sample Run 3" as specified in my code? Sorry if this is super easy, I am very new to JMP/JSL.

 

Thanks

Re: Adding augmented data points to the end of an already existing data table through JSL

I believe this is the most straight forward way to do what you want:

d = DOE( Augment Design, X( :X1 ), Y( :Ystat )/*, {Make Design, Make Table}*/ );
d << Make Design;
dt2 = d << Make Table;
dt << concatenate( dt2, Append to first table( 1 ) );

Remember, JMP isn't Excel or R.  In JMP many times you do not have to write complex code to get things done.  Many times, it is much easier using one of the JMP Platforms, like "concatenate"