MODDE to JMP

[#1]

I have received the DoE model and the results of a study from a partner who uses MODDE ( MODDE® - Design of Experiments Software | Sartorius). Can someone help me how to import existing DoE model and data from MODDE to JMP?

[#2]

 I'm not familiar with MODDE software...but two thoughts:

1. How hard would it be to just recreate the design in JMP natively? Reason I ask is there is loads of meta data that JMP creates for a JMP generated design that can be very, very useful for analysis and reporting/sharing. If the design is a classic design (any 2**k, 2**(k-p) fractional factorial design, n run Plackett - Burman, or RSM oriented design such as a central composite, Box - Behnken, with or without blocking etc.) it is probably worth your while to recreate the design in JMP, then just copy/paste the response data into the JMP generated design.

2. If the design is problematic recreating in JMP...for example a d-optimal design might be difficult to recreate in JMP natively...maybe your path of least resistance is just save the MODDE generated design and responses into a spreadsheet app like Excel...then just copy paste into a JMP data table. You can add some of the meta data field values via JMP Column Properties entries by hand...a royal pain,,,but could be done.

I would definitely choose option 1 above if it doable.

[#3]

Thankyou for your feedback. Its not too complicated to recreate the design by inputting factors etc into JMP. One wrinkle is that  the study was already executed with a specific order of runs based on MODDE design. When JMP generates the DoE runs, they will never match the order of runs executed due to randomization. Is there any option to modify the sequence of DoE runs in the design generated by JMP to match the order already executed?

[#4]

Here's a workaround...admittedly a bit klunky...but I can't think of a way to cajole the randomization sequence created in MODDE to be repeated in JMP. Maybe somebody else knows? So here's how I'd go about getting the MODDE design, with the retained randomization sequence, into a JMP generated design. My thinking is actually a two step merger of my original reply...

Step 1: Create the design in JMP. You still want to do this to retain all the meta data that JMP associates with your factors and responses. You'll want to make sure you have a full understanding of these fields. Things like JMP data type, factor type (blocking, if incorporated into the design,,,) etc. Once created, delete all the rows in the JMP data table...but don't clear out the column headers or any JMP column property data or information.

Step 2: Export the MODDE design and responses, just the actual design, no column headers, or any other information, to say Excel. I'm not sure of the exact format of the 'design' in MODDE. If it's NOT consistent with the format of a JMP generated design, that is each factor has it's own column, each row represents a full treatment combination within the design, then you'll have to manipulate to Excel column by row format to exactly match what JMP would have generated. Make sure the columns in the Excel sheet are in the same sequence left to right as they are in the JMP data table. If you have to export any column headers, delete then from the Excel spreadsheet before proceeding.

Step 3: Copy the entire MODDE design and responses (now in Excel), then do a group paste into the JMP data table.

Step 4: I'd spend a few minutes making sure in the importing process JMP didn't change column properties like JMP data type and modeling type. It's conceivable that say you had a factor called "Supplier" and in the MODDE design they just entered a 'number' like 1, 2, and 3 (for three levels of what we'd call in JMP a categorical factor) and in all the exporting and copy/pasting, JMP kind of 'guesses' the most appropriate JMP data and concurrent modeling type. And depending on what JMP 'sees' when it imports a column may influence what happens. If JMP changes the data type or modeling type to something inappropriate, it's easy enough to change in the column properties menu.

Whew...maybe somebody else has a more elegant solution path?

Once you're done with the above...I'd try and convince whoever created the design in MODDE to get a copy of JMP...maybe they don't know what they are missing?