"Linear Regression with Only Positive Weights"

jw00000

Rapidminer noob here,

I'm interested in performing a Linear Regression for the purposes of making a predictive model, but I would like to force the weights to be positive for each of the attributes. Likewise I may want to force the y-intercept to be zero.  Is this possible?

Thanks,
Jon

jw00000

Crickets.... Anyone else trying to do something similar?  In general I'm interested in bounding the weights by some constraints.  Curiously, if I choose to use a Polynomial Regression it offers me min/max coefficient parameters.  This confuses me since I thought that Polynomial Regression was merely a Linear Regression under the covers.