Subscribe Bookmark RSS Feed

Help with Custom Design

vikram

Community Trekker

Joined:

Apr 27, 2013

Hello,

I am trying to set up a design in the following way:

Custom Design
FactorTypeLevel
Drop HeightDiscrete Numeric3
SizeDiscrete Numeric3
MaterialCategorical2
ReplicatesBlocks3

As mentioned I am using the custom design option, and using 18 runs per block. And I have 3 response variable and a total of 54 total runs. The problem I am facing is that when JMP is making the design table, the distribution of levels is not unique. Sometime times there are more than 3 replicates for a certain level, some level 2 of the discrete numeric factors are getting skipped. I have included the following along with this most:

1) Custom Design - JMP - Full Model: This is the actual project model based on the table mentioned above. The highlighted items are the ones that are getting unnecessarily duplicated

2) Custom Design_Vikram : This is the JMP project file, which I quickly setup to show that level 2 is missing for discrete numeric factors X1 and X2.

Can you please suggest the correct way to deal with this?

1 ACCEPTED SOLUTION

Accepted Solutions
Solution

I'm not sure what is going on. You have not changed the model terms have you? There are some "if possible" terms that are required to drive the model.

11 REPLIES
louv

Staff

Joined:

Jun 23, 2011

Vikram,

I had no problem obtaining a balanced design with your conditions. I just increased the number os starts (under the red triangle) to 1000 and all three of my blocks and also set my user specified run number = 54 and I had a balance of the discrete numeric factors and categorical factors.

vikram

Community Trekker

Joined:

Apr 27, 2013

Thanks for your reply. Did you use the custom design option? One thing I did not do is use the number of starts as 1000, as you mentioned.

vikram

Community Trekker

Joined:

Apr 27, 2013

I changed the start to a 1000, but still I am not getting the correct design. I have copied the custom design tab in this post. If you see for discrete numeric variables X1 and X2 level 2 is missing. Thanks.

Design

Run X1 X2 X3 X4

1 1 1 L2 2

2 1 1 L2 2

3 3 1 L2 3

4 3 1 L2 3

5 3 3 L2 1

6 3 3 L2 1

7 3 1 L2 1

8 3 1 L2 2

9 1 3 L1 2

10 3 1 L2 2

11 3 3 L1 2

12 3 1 L1 2

13 1 1 L2 3

14 1 1 L1 2

15 3 1 L1 3

16 3 1 L1 2

17 1 3 L2 2

18 3 3 L1 1

19 3 1 L1 1

20 3 3 L1 3

21 1 3 L2 1

22 3 3 L2 3

23 1 1 L1 2

24 3 3 L1 2

25 3 1 L1 3

26 3 1 L2 1

27 3 3 L2 2

28 1 1 L1 1

29 3 3 L1 3

30 1 1 L1 3

31 1 3 L1 2

32 1 3 L2 1

33 3 3 L2 3

34 1 3 L1 1

35 1 1 L1 3

36 1 1 L1 1

37 1 3 L2 1

38 1 1 L1 1

39 1 1 L2 1

40 3 1 L2 2

41 1 1 L2 1

42 3 3 L1 2

43 3 1 L1 1

44 1 3 L1 1

45 3 3 L2 2

46 3 3 L1 1

47 1 3 L2 3

48 1 3 L1 3

49 1 3 L2 2

50 1 3 L1 3

51 1 3 L2 3

52 1 3 L1 3

53 1 1 L2 3

54 3 1 L1 3

louv

Staff

Joined:

Jun 23, 2011

Did you specify 18 runs per block?

vikram

Community Trekker

Joined:

Apr 27, 2013

Yes I did. In custom design when I added the blocking factor, I selected custom of 18 runs per block. Then I selected 3 replicates. To make it 3 runs with 18 runs each to include all treatment combinations in one block.

louv

Staff

Joined:

Jun 23, 2011

I don't think that you need to specify the replicates.

enter your factors

enter your block of 18 runs per block

enter 54 as number of runs

vikram

Community Trekker

Joined:

Apr 27, 2013

I tried this just now, following what you mentioned. I did not enter the number of replicates this time. I just entered user defines number of runs as 54. It automatically set 18 runs for each block. But I am still getting the wrong design, where level 2 of discrete factors are missing.

Solution

I'm not sure what is going on. You have not changed the model terms have you? There are some "if possible" terms that are required to drive the model.

vikram

Community Trekker

Joined:

Apr 27, 2013

No I have not changed the model terms. Thanks anyways..