01-23-2015 10:16 AM
In linear regression, can we possibly force the fitting coefficient be less than zero?
I am doing some analysis for price elasticity, a larger-than-zero-coefficient is usually doesn't make sense. This is the reason I am asking if we can force the fitting result of coefficient be less than zero.
01-23-2015 10:36 AM
I think you can do the regression in PROC CALIS (the docs give an example) and use the BOUNDS statement to force the coefficient to be less than zero.