0
$\begingroup$

I am reading a book and have difficulty in understanding the math on bias- variance tradeoff. Below is the section that I am having trouble with:

Given a set of training samples $x_1, x_2, ..., x_n$ and their targets $y_1, y_2, ..., y_n$, we want to find a regression function, $\hat{y}(x)$, which estimates the true relation $y(x)$ as correctly as possible. We measure the error of estimation, how good (or bad) the regression model is by mean squared error ($MSE$): Formula

I can derive mean squared error with partial derivative and the concept of slope. I also understand that $MSE$ is to minimize the total error. I also understand basics statics on expected value.

Yet, I have been stuck in finding the relevant math and statistical concepts behind this formula for a week.

The question is, what are the relevant math and statistical concepts behind this formula?

For example, how

$MSE = E[(y-\hat{y} )^2]$

becomes:

$= E[(y-E[\hat{y} ] + E[\hat{y}] - \hat{y} )^2]$

Thank you! I can see that the first component after adding and subtracting E[y^] is unchanged. Then the formula operated according to $(a+b)^2 = a^2 + 2ab + b^2 $ where

$2ac = +E[2(y - E[\hat{y}])(E[\hat{y}] - \hat{y})]$

Why 2ac becomes

$2(y - E[\hat{y}])(E[\hat{y}] - E[\hat{y}]) $

$\endgroup$
2
  • $\begingroup$ Because adding and subtracting $E[\hat y]$ does not affect the result. $\endgroup$
    – Emre
    Jul 28, 2018 at 18:12
  • $\begingroup$ Thank you. However, I am still having trouble to understand the following maths. I can see that the first component after adding and subtracting E[y^] is unchanged. Then the formula operated according to (a+b)^2 = a^2 + 2ab + b^2 where 2ac = +E[2(y - E[y^])(E[y^] - y^)]. Why 2ac becomes 2(y - E[y^])(E[y^] - E[y^]) ? $\endgroup$
    – Carch
    Jul 29, 2018 at 4:24

1 Answer 1

1
$\begingroup$

Because it's a costant, everything that is a costant value remains unchanged by the E, that's why you can "move" it outside.

For example:

enter image description here

If y is a costant or it is known, the E doesn't affect it, so you can "move" it outside the symbol, and write just y.

The only thing that is unknown is the estimator yhat, in fact you have enter image description here and not just enter image description here.

That's what's happening here.

I found this on CrossValidated that might be more clear.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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