JMP Wish List
We want to hear your ideas for improving JMP. Share them here.What inspired this wish list request? I got into a bug(???) where if a Combined Data Table was made on a collection of plots with various Group By columns, they were lost.
Let's run this code:
Names Default To Here(1);
dt = open("$SAMPLE_DATA/Big Class.jmp");
fg = dt << Fit Group(
Bivariate(
Y( :height ),
X( :weight ),
Group By(:age),
Fit Line
),
Bivariate(
Y( :height ),
X( :weight ),
Group By(:age),
Fit Line
),
Bivariate(
Y( :height ),
X( :weight ),
Group By(:age),
Fit Line
),
Bivariate(
Y( :height ),
X( :weight ),
Group By(:sex),
Fit Line
),
<<{Arrange in Rows( 4 )}
);You'll get a Fit Group where the last Bivariate has a different Group By column (:sex instead of :age)
Then let's click on the first "Summary of Fit" and select Make Combined Data Table.
We'll see that the age column has missing values for the plot that has Group By column NOT :age, and there is no column called "sex"
What is the improvement you would like to see?
Group By is as much of a variable in analysis as X or Y. Why don't we have separate columns for it? Group By Name and Group By Value? This will make automating further analysis much easier.
Worst case scenario, when making combined data table - just use the same logic as concatenating tables with non-matching columns - all columns get in the result with missing values where column did not present in the table.
Why is this idea important? Having Linear Fit or other models and looking at the report is not the end of the analysis - it might be the very beginning. If we have many models, many conditions, many things to look at, robust combined data table would help to automate this