Need help on DOE choice experiment with constraints

unknown

Hi, Your help is really appreciated... Please.
I used a custom design to include constraints with which certain combinations are blocked (e.g., high price and low quality). I collected data. Now I am trying to analze the data. But it keeps giving error that "Cannot find combination 59 in profile data for data row 3. Cannot find combination 79 in profile data for data row 6.
Cannot find combination 52 in profile data for data row 9...." This is my first tme using JMP.

(X1, X2, X3, X4, X5, X6, Y, Profile Number).
Interaction x1x2, x1x4, x1x6
X1 == 5 & X2 == 3 | X1 == 6 & X2 == 3 | X1 == 5 & X6 == 3 | X1 == 6 & X6 == 3

Below are the data.

L2 L2 L1 L3 L3 L2 3 1
L5 L2 L1 L1 L3 L1 5 2
L1 L2 L3 L3 L3 L2 13 3
L1 L3 L3 L1 L1 L1 59 4
L6 L1 L1 L1 L3 L1 10 5
L1 L1 L3 L2 L1 L3 35 6
L4 L3 L3 L2 L3 L2 79 7
L1 L2 L1 L3 L1 L1 21 8
L1 L2 L2 L1 L3 L3 11 9
L2 L3 L2 L2 L1 L2 52 10
L3 L1 L3 L2 L2 L1 43 11
L2 L1 L1 L2 L3 L1 20 12
L4 L2 L3 L1 L3 L1 32 13
L6 L1 L3 L2 L2 L2 43 14
L2 L3 L1 L1 L3 L3 4 15
L2 L2 L1 L2 L3 L3 8 16
L1 L1 L2 L2 L2 L1 20 17
L1 L2 L3 L1 L2 L3 76 18
L2 L1 L2 L3 L1 L3 22 19
L4 L2 L2 L2 L2 L3 33 20
L1 L1 L1 L1 L2 L2 23 21
L2 L2 L2 L1 L1 L1 20 22
L3 L3 L2 L1 L3 L1 16 23
L5 L1 L2 L2 L3 L1 38 24
L5 L2 L3 L2 L2 L2 57 25
L1 L3 L1 L2 L3 L2 23 26
L6 L2 L3 L1 L2 L2 91 27
L5 L1 L1 L1 L3 L2 7 28
L4 L1 L1 L2 L1 L1 23 29
L3 L3 L2 L3 L3 L2 9 30
L4 L3 L1 L1 L1 L3 44 31
L3 L1 L1 L3 L3 L3 7 32
L5 L1 L3 L3 L2 L1 11 33
L4 L1 L2 L1 L2 L2 35 34
L1 L2 L3 L1 L2 L1 43 35
L4 L3 L2 L3 L2 L1 10 36
L3 L3 L3 L2 L2 L3 63 37
L6 L2 L1 L2 L3 L1 24 38
L4 L1 L3 L3 L3 L3 22 39
L3 L2 L3 L3 L2 L1 12 40
L5 L2 L2 L3 L3 L2 8 41
L1 L1 L3 L3 L3 L2 7 42
L2 L1 L3 L1 L2 L2 40 43
L6 L2 L3 L3 L2 L1 24 44
L4 L2 L1 L3 L1 L2 10 45
L3 L2 L3 L1 L2 L2 72 46
L3 L2 L2 L2 L3 L3 20 47
L1 L3 L1 L3 L2 L3 6 48
L2 L2 L2 L1 L1 L2 53 49
L6 L2 L3 L3 L2 L2 24 50
L3 L3 L2 L3 L3 L1 19 51

louv

Hello,

I think the confusion here may be the fact that you attempted to model this custom design using the choice design modeling capability. I believe what you have here is a custom designed experiment with constraints and you need to utilize the Analyze>Fit Model rather than the Analyze>Modeling>Choice.
When I look at your data and just choose a main effects model it appears that if you want to maximize your response the key drivers appear to be D>C>E>B>F with Factor A having little impact. You can also try the Modeling>Screening platform which indicates C = D > (C*D) > E. Both cases indicate an L3 level setting for Factor C coupled with a L1 or L2 level of Factor D are most important if you are trying to maximize your response. You could even take a look at your data with the Analyze> Fit Y by X and visualize the differences using the Oneway output. Hope this helps.