How to combine several linear regression function into one

1.2k Views Asked by At

I have the relationship between f and each of other five variables, the relationships are acquired by linear regression and are as follow

$f = a{\tiny1}x{\tiny1} + b{\tiny1}$

$f = a{\tiny2}x{\tiny2} + b{\tiny2}$

...

$f = a{\tiny5}x{\tiny5} + b{\tiny5}$

I am thinking since the regression is already done and all x variables are related to the same y, and the range of y and x are the same (ranging from 0 to 100). I can just add those five linear functions together and do a normalization or scaling. Then, by feeding the added-up function like

$f = $$\sum_{i=1}^5 (Ai*Xi + Bi)$

with $x{\tiny1}, x{\tiny2}, .., x{\tiny5}$, I can get only single one y value.

Can anyone tell is this way valid, if so, in my case a's range is [0, 1], however, for the intercept b, they are quite large, how can I do the normalization or scaling in order to obtain a valid and reliable value y?

Thanks for any help in advance.

1

There are 1 best solutions below

2
On

It depends on the structure of your data. I.e., if you can view it as $n$ vectors of a kind $$ (y_i, x_{1i}, x_{2i}, ..., x_{ni}), \quad i=1,...,n $$ then a multiple linear model should fit to the task, namely $$ y=\beta_0 + \sum_{j=1}^5\beta_jx_j + \epsilon. $$ That you can estimate using any statistical software.