BookmarkSubscribeSubscribe to RSS Feed
rsgregorian

New Contributor

Joined:

Jun 13, 2018

A custom design with all categorical factors

Hi Everyone.  

I'm a little bit of a neophyte here.  So bear with me.  

I'm trying to design an experience with 4 categorical variables and no continuous variables.  I have 2x2level variables 1x3 level and 1x6 level.  I could devolve that one into 1x3 and 1x2 if need be.  

 

I've gone through custom design and everything seems to make sense, but all of the D efficiency etc statistics are blank.  I am needing 18 runs, and I thought I could block that into two groups of 9.   What am i missing here?  Does this signal a warning I need to worry about?

 

Thanks for your help  

 

3 REPLIES
markbailey

Staff

Joined:

Jun 23, 2011

Re: A custom design with all categorical factors

This script will produce the design that you described:

DOE(
	Custom Design,
	{Add Response( Maximize, "Y", ., ., . ),
	Add Factor( Categorical, {"L1", "L2"}, "X1", 0 ),
	Add Factor( Categorical, {"L1", "L2"}, "X2", 0 ),
	Add Factor( Categorical, {"L1", "L2", "L3"}, "X3", 0 ),
	Add Factor( Categorical, {"L1", "L2", "L3", "L4", "L5", "L6"}, "X4", 0 ),
	Add Factor( Blocking, 9, "X5" ), Set Random Seed( 6155315 ),
	Number of Starts( 64566 ), Add Term( {1, 0} ), Add Term( {1, 1} ),
	Add Term( {2, 1} ), Add Term( {3, 1} ), Add Term( {4, 1} ), Add Term( {5, 1} ),
	Add Alias Term( {1, 1}, {2, 1} ), Add Alias Term( {1, 1}, {3, 1} ),
	Add Alias Term( {1, 1}, {4, 1} ), Add Alias Term( {2, 1}, {3, 1} ),
	Add Alias Term( {2, 1}, {4, 1} ), Add Alias Term( {3, 1}, {4, 1} ),
	Set Sample Size( 18 ), Simulate Responses( 0 ), Save X Matrix( 0 ), Make Design}
);

It does not exhibit the problem you described.

 

Learn it once, use it forever!
rsgregorian

New Contributor

Joined:

Jun 13, 2018

Re: A custom design with all categorical factors

Thank you mark.  I think i made the mistake of not declaring the blocking variable.  This makes much more sense.  Still learning. 

markbailey

Staff

Joined:

Jun 23, 2011

Re: A custom design with all categorical factors

Well, it depends...

One of my (much smarter than I am) colleagues in JMP Technical Support pointed out that if you chose to block your runs for a random blocking effect, then the efficiencies are not computable and, therefore, not reported. I assumed that you entered a blocking factor because you expected a fixed blocking effect. (Why did I assume that? I don't really know and now I have no good reason.) JMP allows you to block your runs for either kind of effect (but not both) so it just depends on how you think about the effect of the blocks (contributed variation) on the response and, therefore, how you will model the variation from the blocks.

  • If you think that the blocks have a fixed effect on the response, then add a blocking factor. (A term to estimate the fixed effect is automatically added to the model.)
  • If you think that the blocks have a random effect on the response, then do not add a blocking factor but select to group the runs at the bottom of Custom Design.
Learn it once, use it forever!