How to estimate the constant term in a multiple linear regression?

Hi,

I would like to estimate the constant term "a" in the multiple linear regression X=a+bY+BZ

This is my code :

y =  [Y Z]
[h(1).mdl]=mvregress(X,y,'algorithm','ecm')

Answer 1

[the cyclist]

Add a constant term to your predictor variable:

y = [ones(size(Y)) Y Z]

[Mr LE]

I try this but I obtain an implausible constant term ...

Answer 2

[the cyclist]

Can you post a very small example where you are seeing implausible results? The syntax of mvregress can be a bit tricky (and I actually think my simple suggestion may actually a bit misleading.)

I have posted a very detailed example of using mvregress in this question that you might want to take a look at.

There is also useful information in this question.

Answer 3

[Tom Lane]

You write "X=a+bY+BZ" suggesting that X may be a response and Y and Z predictors. But the first input to mvregress is the predictor matrix, and the second is the response matrix. Perhaps you need to swap them, or maybe you just need to explain what you're trying to fit.