947936_en-US

liliconcom

How can I recreate a lost screening DOE design

I created a classical screening DOE with 6 factors. Using the default number of runs of 20 the jmp software made a design table. I didn't save the design table. I ran the DOE and collected the responses. To analyze the DOE and create prediction formulas I need the original design table. Each time I set up a screening DOE with the same factors and responses I get different design tables. Is there a way to recreate the original design table?

Design of ExperimentsRe: How can I recreate a lost screening DOE design

Are you sure it is not the same design? JMP randomizes the runs by default each time, but it might be the same set of runs.

Also, how did you run the experiment without the JMP data table? Did you export it to a spreadsheet to collect the data? If so, you can import the spreadsheet into JMP. With that done, you just have t add a few column properties to proceed.

Re: How can I recreate a lost screening DOE design

I compared the two tables very closely. Seven out of twenty runs are the same the other 13 are different.

I took a screenshot of the design to paste into a powerpoint to capture the design details. My computer shut down for a scheduled update when I went home and closed jmp. Unfortunately I didn't save the table. However, I can recreate an excel spreadsheet to match the original design.

 

Re: How can I recreate a lost screening DOE design

if you have the experimental design table in excel this should work.
1. Go into the DOE platform and recreate your experiment, same factors and responses

2. after you make the table, select all the rows and use Clear to remove the values.

3. Assuming the columns are in the same order, copy the values from excel and paste them into the JMP table.

-this sets you up to have all the table scripts and column metadata (properties) set up correctly. 

 

Re: How can I recreate a lost screening DOE design

Cool work around that I hadn't thought of for the Excel paste method. I was wondering about column properties and metadata and the scripts being a nightmare to recreate from scratch manually in a virgin JMP data table.