Constrained Multiple Linear Regression using lsqlin

Hi All,

I am trying to solve a multiple linear regression equation: y = a*x1 + b*x2 + c*x3 + d*x4 + e

where x's are my inputted data (all of same length), and I am solving for a-e. Originally, I successfully used 'regress' to obtain by solutions, but I realized after that I needed to constrain a and b to be >= 0. I am trying to transition from 'regress' to 'lsqlin' (I have the optimization toolbox), but the syntax for 'lsqlin' is confusing me and my answers aren't making sense. Could someone please help check my input parameters? I think I'm almost there but I am missing something, primarily with my "d" vector (I have no idea what I should have here, i.e., how the constant vector works for multiple linear regression cases). Thanks so much!!

C = [blankF NPF wavelength.^-4 wavelength ones(size(wavelength))];

d = [ones(size(wavelength))];

lb = [0 0 -Inf -Inf -Inf];

x = lsqlin(C,d,[],[],[],[],lb,[])

Answer 1

[Hannah Doss]

For anyone who was also confused by this, I solved my own problem! The d vector is your y.