changed behavior in JM19:
https://community.jmp.com/t5/JMPer-Cable/New-in-Graph-Builder-for-JMP-19/bc-p/904473/highlight/true#M1127
It's great that new functions are being added. But, JMP team, please (!!!) keep the old functionality as the default setting!
Otherwise, users will have to change hundreds of scripts!
We wondered about the new behavior:
it always worked. And "today" -- why are the main groups split into small chunks ?
@shannon-c-b , thanks for pointing to the root cause of the "bug":
As a workaround, one can create an auxiliary column to simulate the default (JMP ?-18) functionality of order by () with JMP19:
dt = Open( "$SAMPLE_DATA/Big Class.jmp" );
New Column( "myOrderBy",
Formula( Col Mean( :height, :age ) * 10000000 + :height ),
) << run formuals();
Graph Builder(
Variables(
X( :age, Order By( :myOrderBy, Ascending, Order Statistic( "Mean" ) ) ),
X(
:age_name,
Position( 1 ),
Order By( :myOrderBy, Ascending, Order Statistic( "Mean" ) )
),
Y( :height )
),
Elements( Bar( X( 1 ), X( 2 ), Y ) )
);
Hi @shannon-c-b
important: my workaround is not an exact 1:1 copy of the JMP18 style.
[TS-00240917]
my workaround:
- sorts the ages by mean(height)
- sorts the students by height within the age groups
JMP < v19:
- sorts the students by mean(height) -> stores the order (*)
- sorts the ages by mean(height)
- places the students into the respective age groups - position by (*)
as Robert shows up twice, Robert is positioned by mean(height)
-> Please note the jumps around Robert #1 and Robert#2
my application case is for "students" with unique names, this is why I did not notice the difference.
If your dataset contains "students" with duplicate names and the above workaround is not suitable for you (i.e. if you require the specialty (*) ), please contact JMP Support (referencing TS-00240917). They will be able to provide a workaround.
subcategories sorted by order within the parent category?
-> here is the wish: options for "order by" in Graph Builder
