Multiple linear regression with constraint

 I need to run a multiple linear regression such that all of the coefficients are positive, and their sum = 1 . I have 7 variables, so i need 7 seven coefficients.
Thanks !

Answer 1

[Amit]

Use lsqlin ( http://www.mathworks.com/help/optim/ug/lsqlin.html )

[abdoulaye thiam]

Thanks Amit, with lsqlin i am able to set the sum of the coefficients being equal to zero, but i do not know how to make sure that they are all positive numbers.. sorry, if the question sounds dumb, i am new matlab user thank

[Amit]

make sure that all the coefficients are positive, you set the lower bound as 0 and upper bound as inf. something like:

lsqlin(.......,zeros(7,1),Inf*ones(7,1))

[abdoulaye thiam]

Hello Amit, even after setting the lower bound to zero it gives negative coefficients.

[Amit]

How did you set it? How did you use lsqlin?

[John D'Errico]

Note that most likely, some of the coefficients MIGHT go negative by an amount on the order of eps. Look at the numbers. If they are -1.e-13, why do you care? Just set them to ZERO afterwards.

[abdoulaye thiam]

Yes you are right John, the only the only negative number that i have is -2.641592169276901e-17...Sure, why would I care!! Thank you both!!