Choose Language Hide Translation Bar
Highlighted
khern
Level I

How to Constraint Factors to not be specific states

Hello all,

I am a novice and I am actually trying to set up my first DOE using JMP.

I would like to setup a couple of constraints for two of my factors and I cannot get it to work.

I have 3 factors where x1 can only be 0 and 1, while x2 and x3 can be -1,0,1.

 

*The first constraint is when x1 is 0, x2 and x3 can only be -1 or 0.

 Similarly, when x1 is 1, then x2 and x3 can only be -1 or 1.

 

*The other constraint is x2 and x3  cannot be -1 at the same time, regardless of what x1 is.

 I feel like the logic is straight forward but I do not know how the script coding works and the logic structure I put in doesn't compile.

 

Any help would greatly be appreciated; if this is even possible in JMP

 

Regards,

KVH

 

 

1 ACCEPTED SOLUTION

Accepted Solutions
Highlighted

Re: How to Constraint Factors to not be specific states

The means of defining constraints depends on the type of the factors involved. You talk about specific levels so it seems you are thinking about categorical factors and disallowed combinations.

 

Also, adding constraints to the factor combinations will limit the effects that may be estimated.

 

I used the disallowed combinations filter in Custom Design to specify your constraints. It looks like this picture when set up correctly:

 

Screen Shot 2020-08-29 at 8.04.58 AM.png

 

The result is this D-optimal design in 18 runs:

 

Screen Shot 2020-08-29 at 8.07.03 AM.png

 

I include a script below that will produce the same Custom Design platform so that you can modify anything you like:

 

DOE(
	Custom Design,
	{Add Response( Maximize, "Y", ., ., . ),
	Add Factor( Categorical, {"0", "1"}, "X1", 0 ),
	Add Factor( Categorical, {"-1", "0", "1"}, "X2", 0 ),
	Add Factor( Categorical, {"-1", "0", "1"}, "X3", 0 ),
	Set Random Seed( 281228126 ), Number of Starts( 12705 ), Add Term( {1, 0} ),
	Add Term( {1, 1} ), Add Term( {2, 1} ), Add Term( {3, 1} ),
	Add Alias Term( {1, 1}, {2, 1} ), Add Alias Term( {1, 1}, {3, 1} ),
	Add Alias Term( {2, 1}, {3, 1} ), Set Sample Size( 18 ),
	Disallowed Combinations(
		X1 == "0" & X2 == "1" | X1 == "0" & X3 == "1" | X1 == "1" & X2 == "0" | X1
		 == "1" & X3 == "0" | X2 == "-1" & X3 == "-1"
	), Simulate Responses( 0 ), Save X Matrix( 0 )}
);
Learn it once, use it forever!

View solution in original post

1 REPLY 1
Highlighted

Re: How to Constraint Factors to not be specific states

The means of defining constraints depends on the type of the factors involved. You talk about specific levels so it seems you are thinking about categorical factors and disallowed combinations.

 

Also, adding constraints to the factor combinations will limit the effects that may be estimated.

 

I used the disallowed combinations filter in Custom Design to specify your constraints. It looks like this picture when set up correctly:

 

Screen Shot 2020-08-29 at 8.04.58 AM.png

 

The result is this D-optimal design in 18 runs:

 

Screen Shot 2020-08-29 at 8.07.03 AM.png

 

I include a script below that will produce the same Custom Design platform so that you can modify anything you like:

 

DOE(
	Custom Design,
	{Add Response( Maximize, "Y", ., ., . ),
	Add Factor( Categorical, {"0", "1"}, "X1", 0 ),
	Add Factor( Categorical, {"-1", "0", "1"}, "X2", 0 ),
	Add Factor( Categorical, {"-1", "0", "1"}, "X3", 0 ),
	Set Random Seed( 281228126 ), Number of Starts( 12705 ), Add Term( {1, 0} ),
	Add Term( {1, 1} ), Add Term( {2, 1} ), Add Term( {3, 1} ),
	Add Alias Term( {1, 1}, {2, 1} ), Add Alias Term( {1, 1}, {3, 1} ),
	Add Alias Term( {2, 1}, {3, 1} ), Set Sample Size( 18 ),
	Disallowed Combinations(
		X1 == "0" & X2 == "1" | X1 == "0" & X3 == "1" | X1 == "1" & X2 == "0" | X1
		 == "1" & X3 == "0" | X2 == "-1" & X3 == "-1"
	), Simulate Responses( 0 ), Save X Matrix( 0 )}
);
Learn it once, use it forever!

View solution in original post

Article Labels