0
$\begingroup$

My question is close to this one except for one term. More specifically:

$$\frac{1}{m}\sum_{i = 1}^m (x_i - \mu)^2$$

If I wanted to find the derivative of this entire term w.r.t. $x_i - \mu$, how would I go about that? Is this even possible?

Edit

I've tried the derivation but I'm not sure if this is correct. I also left out an important detail that in this context $\mu = \frac{1}{m}\sum_{i = 1}^m x_i$ (i.e. the mean of $x$'s).

$$ \begin{align} \frac{\partial \sum}{\partial (x_i - \mu)} (\frac{1}{m}\sum_{i = 1}^m(x_i - \mu)^2) & = \frac{\partial \sum}{\partial (x_i - \mu)} \frac{1}{m}((x_1 - \mu)^2 + \cdots + (x_m - \mu)^2) \\ & = \frac{1}{m}(2(x_1 - \mu) + \cdots + 2(x_m - \mu)) \\ & = \frac{2}{m}( (x_1 + \cdots + x_m) - 2m\mu) \\ & = 2 \times \frac{1}{m}\sum_{i = 1}^m x_i - 4\mu \\ & = 2\mu - 4\mu \\ & = -2\mu \end{align} $$

I'm not entirely sure if this derivation is correct...

$\endgroup$
0
$\begingroup$
  • Rename your variables (for example $t_i = x_i - \mu$). So yes, you can differentiate with respect to $t_i$.
  • Result $$ \frac{2}{m} \left( x_i - \mu \right) $$
$\endgroup$

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.