How do I graph the desirability sweet spot against factor levels

[#1]

Hi All,

I've found JMP great to find out what the optimal settings are in a design, but I'm having some trouble creating a visualisation of what the sweet spot looks like for a given design. 

I have an example with a JMP file, attached, which I can share. This is a DOE on optimising tea. The desirability settings are:

Factor	Low	High
Sweetness	1.5	2.5
Flavour	3	4.5
Colour	0.35	0.45
Temperature	3.8	4.5

Another DOE software can make a plot like the one below. I find this graph quite compelling because you can easily tell the size of the sweet at different factor levels, which might be desirable for robustness.

I've tried to use the contour plot to visualise the space, but a limitation appears to be that I can't show how the sweet spot changes as the sugar level changes in one graph. You can easily tell how large the sweet spot area is at different factor levels. I'm optimistic that I can reproduce this graph in graph builder but I'm stuck at inputting the desirability and confidence interval into it (or maybe I need to try a different approach)

What is the best way to approach this?

[#2 · Accepted solution]

There could be a better way but I often resort to simulating the results to a new table and graphing from that.  I don't know of a way to directly draw a color or contour in graph builder from a formula (maybe someone can show me up here).  

What I did:

• Open profiler with pred sweetness as Y
• From red triangle use 'reset factor grid' to pick bounds for each variable
• From red triangle pick output random table - choose a lot of rows if you want smooth graphs
• Draw charts using the new data table

This feels like a brute force method, maybe there is something better out there...

[#3]

Thank you @ih . I've had a go at building on this approach. I've never tried outputting a random table before.

The plot you showed has sweetness as the Y. I've replotted it with Sweetness desirability as y, which shows where best sweetness is, not just low to high. It makes sense, too much sugar is too sweet. Too little has no flavour.

I didn't give enough information in the first email. The other responses are colour, flavour and temperature as well, and I need to visualise the intersecting sweetspot of all these responses to result an "in-spec" tea. It's father's day in Australia tomorrow so I need to get this right.

By playing with the prediction profiler, it looks like if desirability >0 implies that that the response is within the lower and upper bounds that I set. The closer to the middle target, the closer it is to the target value in the range.

Based on this I applied a data filter to Sweetness Desirability/Flavour Desirability/Colour Desirability/Temperature Desirability and ommitted all values <0.1 for each response.

I then created a new column where I multiplied each of these desirabilities together to find the best results (assuming equal weighting). The output looked like this:

Do you see any risks with this approach?

@Mark_Bailey I will try this out as well. I have not used the simulate feature before.

[#4]

How to define most desirable from the different components, for example whether equal weighting is appropriate, is a more difficult question that I can't answer, but the method you used sounds logical to me.

Now I might need to up my team game a bit...

[#5]

Expanding on @ih's suggestion, have you used the Simulate feature in the Prediction Profiler?