JSL - How to iterate over Bivariate and save all splines to 1 column

adjo9835

Hello.  I have created an array of strings, containing the various values found in one of my columns. I would like to generate a spline for every value that is in my array of strings, and save all predicted points to a new column.  I am able to iterate over Bivariate(), changing a where-clause with each iteration to a new value in my array of strings, however, only the last predicted points are saved to a new column.  How do I get all predicted points to be saved into a new column?  Thanks.

Proc = "new";
processArray = {};

DataCV=open("C:\Desktop\JMP\sample_data.csv");  

for (l = 1, l < 32521, l++,		// Generate array of strings
	if (proc == (Row()=l; :process), 
		print("same process"), 
		Insert Into(processArray, (Row()=l; :process)); proc = (Row()=l; :process); Show(proc)
	);
);

for (n = 1, n <= length(processArray), n++,	// Generate and save multiple predicteds
	biv = Bivariate(
		Y( :c_mm_deembed_pf ),
		X( :v_dut_volt ),
		Where( :c_type == "Cdg" & :process == processArray[n] ),// :process == processArray[n] appears to be working, as 6 splines are generated
		Fit Spline( 0.1, standardized, {save predicteds})	// {save predicteds} is not saving all 6 splines, only last spline
	);
);

 

txnelson

Here is an annotated reworked version of your script.  Try it out.  I believe it will work with your data.  I was not able to do a real good job of testing, but any errors should be minor

Names Default To Here( 1 );
Proc = "new";
processArray = {};

DataCV = Open( "C:\Desktop\JMP\sample_data.csv" ); 


// In the commented area is a reworked version of your code that I believe
// simplifies the syntax of what you were doing.  Please review it and see 
// what changes I made.  
/*for (l = 1, l <= N Rows(DataCV) , l++,		// Generate array of strings
	if (proc == :process[l], 
		print("same process")
		, 
		Insert Into(processArray, :process[l]); 
		proc = :process[l]; 
		Show(proc)
	);
);*/ 

// The summarize() function is a much faster way to get the values found for 
// a spcific column or combination of columns
Summarize( DataCV, processArray = By( :process ) );

For( n = 1, n <= Length( processArray ), n++, 	// Generate and save multiple predicteds
	// I found that using a JSL variable in the where clause was having issues, because the formula
	// created for the Predicted values contains a reference to PredictedArray, which JSL Variable can change,
	// which makes the formula invalid.  So what needs to be done, is to subsitute the actual value for 
	// predictedArray[n] into the Where Clause
	Eval(
		Substitute(
				Expr(
					biv = Bivariate(
						Y( :c_mm_deembed_pf ),
						X( :v_dut_volt ),
						Where( :c_type == "Cdg" & :process == __process__ ), // :process == processArray[n] appears to be working, as 6 splines are generated
						Fit Spline( 0.1, standardized, {save predicteds} )	// {save predicteds} is not saving all 6 splines, only last spline
					)
				),
			Expr( __process__ ), processArray[n]
		)
	)
);