Proper Way to Use JSL Derivative

[#1]

I have been attempting to create Derivatives of a Response Surface Model with JSL for some time now. The lack of documentation and forum posts here only pointing toward the manual creation of derivatives have not solved my issue. I already have a Fit, and the Prediction Formula saved to a column of the data table. What I'd like to do is use JSL to create the symbolic derivative of the formula and apply it to another column. So far the following code are all the different ways I have attempted to solve this issue to no avail.

// Get Formula
//formula = col << Get Formula;
formula = Expr((-211374703.282551) + 30.8532207444449 * :X1);
Show( formula );
// Make derivative
derivative = Derivative(formula, X1);
Show(derivative);
derivative = Derivative(formula, :X1);
Show(derivative);
derivative = Derivative(formula, "X1");
Show(derivative);
derivative = Derivative(formula, ":X1");
Show(derivative);
derivative = Derivative(formula, Column(:X1));
Show(derivative);
derivative = Derivative(Eval(formula), X1);
Show(derivative);
derivative = Derivative(formula, Eval(X1));
Show(derivative);
derivative = Derivative(Eval(formula), Eval(X1));
Show(derivative);
derivative = Derivative(Expr(formula), X1);
Show(derivative);
derivative = Derivative(formula, Expr(X1));
Show(derivative);
derivative = Derivative(Expr(formula), Expr(X1));
Show(derivative);

And the output I get is

formula = (-211374703.282551) + 30.8532207444449 * :X;
derivative = 0;
derivative = 0;
derivative = .;
derivative = .;
derivative = .;
derivative = 0;
derivative = .;
derivative = .;
derivative = 0;
derivative = .;
derivative = .;

Typically the formula is being extracted from the column holding the saved prediction formula, and in this case the printout does successfully show the formula.

What am I missing here? I thought this would be a straight-forward process.

Thanks for any help you might be able to give!

[#2]

Try adding extra Expr to your formula:

formula = Expr(Expr((-211374703.282551) + 30.8532207444449 * :X1));

[#3 · Accepted solution]

formula = Expr((-211374703.282551) + 30.8532207444449 * :X1);
Show( formula );
// Make derivative
derivative = Derivative(name expr(formula), X1);
show(derivative);

The key is to avoid formula being evaluated when it is passed into Derivative. Wrapping a second layer of Expr creates desired behavior but actually peels off the first layer of Expr when it is passed into the call.

[#4]

Thank you for the reply!

Can we take this one step further?

Let's say I have a multivariate function, and I want to loop over a list of inputs that I know beforehand to make derivatives.

So instead the setup would be

formula = Expr((-211374703.282551) + 30.8532207444449 * :MyVar1 - 0.42 * :AnotherVar2);factors = List(MyVar1, AnotherVar2);
// Clone formula n times into columns
dt << Clone Formula Column(
	"Pred Formula",
	N Items(factors),
);
For(i = 1, i < N Items(factors), i++,
	col = Column( dt, "Pred Formula " || char(eval(i+1)) );
	col << Set Name( "doptval d" || factors[i]);
	// Get Formula
	formula = col << Get Formula;
	// Make derivative
	derivative = Derivative(name expr(formula), factors[i]);
	// Apply derivative to new column
	col << Set Formula( derivative );
);

 Now trying to reference the variable in this loop is causing the Derivative function to fail. Any thoughts?