I was interested in seeing how the normal equations for least squares for linear regression are derived, and found this page: https://eli.thegreenplace.net/2014/derivation-of-the-normal-equation-for-linear-regression/

I am fine with most of the derivation except for two key lines. They derive the loss function as: enter image description here

And then simplify this to: enter image description here

I don't understand how $(X\theta)^T y - y^T (X\theta)$ simplifies to $-2(X\theta)^T y$.


Notice that $-(X\theta)^T y = -(X\theta) \cdot y$ where $\cdot$ is the dot product. Similarly, $-y^T(X\theta) = -y \cdot (X\theta) = -(X\theta) \cdot y$ (since the dot product is symmetric).

Hence, the sum of these two values is $-2(X\theta)\cdot y = -2(X\theta)^Ty$


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.