What inspired this wish list request? When performing a DOE with competing responses, user is able to specify the Response Goal importance for each response. Often this is a suggested multiplier. For instance, comparing Brand gasoline and Octane in the early JMP class for Custom DOE, Mileage (mpg) may be 2X the importance of Price. Since the desirabilities are competing, an optimal Octane is obtained around 90.4 or 90.5 (depending on random simulation).
Under Prediction Profiler, for the user to add to "Remembered Settings" table, have to change the Setting name to some code like "mpg 2x of price". Then if the user wants to see if the mpg was 3X the importance of price, change the Variable Importance to 3, and "Maximize and Remember". Once again, have to add a code like "mpg 3x of price" to the Setting name. Would be nice to have the columns added for each Response's importance.
Wish is to have columns for importance added to this table.
What is the improvement you would like to see? Under the Remembered Settings table, include columns for each response's Importance, to track any changes.
An even better improvement to allow user to simulate importance values for a certain range. I did a script for each response, changing the importance from 1 to 5, just to compare.
For(i=1, i<=5,i++,
For(k=1,k<=5,k++,
// Change column property: Mileage (mpg)
Data Table( "Gas Mileage DOE" ):"Mileage (mpg)"n <<
Set Property("Response Limits",{Goal( Maximize ), Lower( 25 ), Upper( 30 ), Importance( i ), Show Limits( 0 )});
// Change column property: Price ($/gal)
Data Table( "Gas Mileage DOE" ):"Price ($/gal)"n <<
Set Property("Response Limits",{Goal( Minimize ), Importance( k ), Show Limits( 0 )});
Why is this idea important? This would further enhance the use of the "Remembered Settings" table under the Prediction Profiler for Fit Model or Fit Least Squares platform. Helps track manually changing the importance and tracking in the "Remembered Settings" table. Allowing a short script to iterate through the responses' importance would allow easier "what-if" scenarios.