Subscribe Bookmark RSS Feed

How do I create a Split Plot RSM with 2 HTC Factors 1 ETC factor 10 whole plots 30 runs with 3 Blocks?

recremekim

Community Trekker

Joined:

May 15, 2015

I can create a RSM Split Plot with 2 HTC and 1 ETC factors easy enough but when I add a blocking factor it doesn't meet the split plot design?

1 ACCEPTED SOLUTION

Accepted Solutions
ryan_lekivetz

Joined:

Nov 1, 2013

Solution

Were you wanting the HTC factors nested within the blocks?

If so, you can try something similar to: Constructing a custom design with two nested random blocking

You would then want to remove all the "false VHTC factor" terms from the model if you use RSM. I've included code below to give you an idea of the setup.

The "Whole Plots" column is your block, while Subplots for your HTC factors. You would remove the "false VHTC factor column". When modeling, you would have to consider switching the blocking factor (ie. Whole Plots) from a random effect.

Hope this helps.

Cheers,

Ryan

DOE(

 

      Custom Design,

 

      {Add Response( Maximize, "Y", ., ., . ),

 

      Add Factor( Continuous, -1, 1, "false VHTC factor", 2 ),

 

      Add Factor( Continuous, -1, 1, "X2", 1 ),

 

      Add Factor( Continuous, -1, 1, "X3", 1 ),

 

      Add Factor( Continuous, -1, 1, "X4", 0 ), Set Random Seed( 1390686764 ),

 

      Number of Starts( 2000 ), Add Term( {1, 0} ), Add Term( {2, 1} ),

 

      Add Term( {3, 1} ), Add Term( {4, 1} ), Add Term( {2, 2} ),

 

      Add Term( {2, 1}, {3, 1} ), Add Term( {3, 2} ), Add Term( {2, 1}, {4, 1} ),

 

      Add Term( {3, 1}, {4, 1} ), Add Term( {4, 2} ), Set N Whole Plots( 3 ),

 

      Set N Subplots( 10 ), Set Sample Size( 30 ), Optimality Criterion( 2 )}

 

);

 

2 REPLIES
ryan_lekivetz

Joined:

Nov 1, 2013

Solution

Were you wanting the HTC factors nested within the blocks?

If so, you can try something similar to: Constructing a custom design with two nested random blocking

You would then want to remove all the "false VHTC factor" terms from the model if you use RSM. I've included code below to give you an idea of the setup.

The "Whole Plots" column is your block, while Subplots for your HTC factors. You would remove the "false VHTC factor column". When modeling, you would have to consider switching the blocking factor (ie. Whole Plots) from a random effect.

Hope this helps.

Cheers,

Ryan

DOE(

 

      Custom Design,

 

      {Add Response( Maximize, "Y", ., ., . ),

 

      Add Factor( Continuous, -1, 1, "false VHTC factor", 2 ),

 

      Add Factor( Continuous, -1, 1, "X2", 1 ),

 

      Add Factor( Continuous, -1, 1, "X3", 1 ),

 

      Add Factor( Continuous, -1, 1, "X4", 0 ), Set Random Seed( 1390686764 ),

 

      Number of Starts( 2000 ), Add Term( {1, 0} ), Add Term( {2, 1} ),

 

      Add Term( {3, 1} ), Add Term( {4, 1} ), Add Term( {2, 2} ),

 

      Add Term( {2, 1}, {3, 1} ), Add Term( {3, 2} ), Add Term( {2, 1}, {4, 1} ),

 

      Add Term( {3, 1}, {4, 1} ), Add Term( {4, 2} ), Set N Whole Plots( 3 ),

 

      Set N Subplots( 10 ), Set Sample Size( 30 ), Optimality Criterion( 2 )}

 

);

 

recremekim

Community Trekker

Joined:

May 15, 2015

Thanks this was right on!

Recrem Ekim

is

M i k e M e r c e r

Spelled backwards

<http://www.linkedin.com/in/MichaelCMercer> www.LinkedIn.com/in/MichaelCMercer