0
$\begingroup$

What does it mean: to force a line through the origin?

I interpret this to be making (forcing) the intercept as (0,0) in the regression procedure of an x-y scatterplot, but am not sure.

$\endgroup$
  • 1
    $\begingroup$ Can you provide more context, for example regarding where someone may have used this expression? $\endgroup$ – Wolfgang Bangerth Feb 11 '16 at 6:20
2
$\begingroup$

I think that what you are trying to do is to find a line passing through a set of data which is able to best fit that set of data. A Least Square approach, for instance, could be used for this purpose. Force a line to pass through the origin means that if you have the general affine equation for the line as

$$ y = mx +q $$

than the affine term $q$ is null, $q=0$.

Let assume that you have a set of data $Y \in \mathbb{R}^{N}$ evaluated in $X \in \mathbb{R}^{N}$. You want to find a relation between these dataset using a linear relation. You can describe this relation using a linear regression of the form

$$ Y = \Phi \Theta $$

where in general $\Theta \in \mathbb{R}^{N+1}$ is the set of unknowns parameters able to desribe the relation between $X$ and $Y$ while $\Phi \in \mathbb{R}^{N \times N+1}$ is a suitable matrix build starting from $X$.

Let assume that $X$ is defined as follows:

$$ X = [ x_1\ x_2\ \cdots\ x_N] $$

and the matrix $\Phi$ as

$$ \Phi = \left[ \begin{matrix} x_1 & x_2 & \cdots & x_N & 1 \\ x_1 & x_2 & \cdots & x_N & 1 \\ \vdots & \cdots & \cdots & \ddots & \vdots \\ x_1 & x_2 & \cdots & x_N & 1 \\ \end{matrix} \right] $$

as it is possible to see, the last column of $\Phi$ multiplies the last element of the unknowns $\Theta$, which is considered as an affine term (beacause of the $1$ value).

If you want to force the regression line to pass through the origin, just set to zero the last column or neglect it while building the $\Phi$ matrix (be careful because if you remove the last column of $\Phi$ then you have also to reduce the size of $\Theta$).

Once you have your $\Phi$ matrix you can solve a linear regression problem (for instance) with the Least Square approach. The objective is to find $\Theta$ able to minimize the following fitness function $$ \|Y-\Phi \Theta\|_2^2 $$ where the operator $\| \cdot \|_2$ represents the two-norm operator of an array.

If no constraints are set, the analytical solution of $\Theta$ is:

$$ \Theta = (\Phi^{T}\Phi)^{-1}\Phi^{T} Y $$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.