What inspired this Wish List Request? This is not a request from customers, but it is just based on my experience in using some modeling platforms.
Please describe the new or improved feature or solution you would like to see.
If we have 0-1 binary columns, it almost always means 0 is the control group or a non-event, and 1 is the treatment group or an event.
When we fit Logistic Model for 0-1 binary columns, 1 is treated as an event at the default. It is OK.
And when we fit JMP Pro 19's Causal Treatment, 1 is treated as the treatment group, and 0 is treated as the control group. It is also OK.
But in many other platforms, 0 is treated as the treatment, and 1 is treated as the control at the default. For example, please run the script below.
dt = New Table( "Test",
New Column( "Treatment", Values( [1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0] )),
New Column( "Covariate", Values( [15, 31, 21, 41, 52, 31, 25, 31, 45, 32, 14, 32, 13] )),
New Column( "Response", Values( [15, 21, 34, 51, 22, 41, 25, 31, 2, 4, 5, 3, 1] ))
);
dt << Fit Model(
Y( :Response ),
Effects( :Treatment, :Covariate ),
Personality( "Generalized Linear Mixed Model" ),
Generalized Distribution( "Normal" ),
Run( Fit( Random Effects Covariance Parameter Estimates( 0 ) ) )
);
We can change the default behavior by Value Ordering column property, but if we do so, Y in Logistic and JMP Pro 19's Causal Treatment does not behave what we usually want.
I am not how we can change, but I feel some inconvenience for the current behavior.
What would the value be to you and/or other users if this new feature is provided? I have no idea.
Is this an external customer request? No.
Is there a competitive software package that has this feature or solution that we should be aware of? I think in many R packages, 1 is treated as the treatment, and 0 is treated as the control for 0-1 binary factor variables.
Will this impact a renewal, expansion, or adoption at a customer, and does this put an account at risk if we do not address this? No.
