The variance is the $2^{nd}$ [[Moments of a Distribution#^afe9bc|central moment]] and takes the following from.
$
\begin{align}
\mathrm{Var}(X) &= \mathbb E \big[(X- \mathbb E[X])^2\big] \\[2pt]
\sigma_X &= \sqrt{\mathrm{Var}(X)}
\end{align}
$
> [!note:]
> - Because of the squared function the variance is always $\ge0$.
> - The standard deviation $\sigma$ takes the square root of the variance, and can therefore be interpreted in original units of $X$.
There are two different ways to compute the variance of a r.v. $X$.
- *Expected value rule:* We can calculate variance with the expected value rule, where we insert the function $g(Z)=(z-\mu)^2$, and calculate the [[Expectation]] of the function.
$
\begin{align}
\mathrm{Var}(X) &= \mathbb E[g(Z)] \\[8pt]
&= \sum_z g(Z)*p_z(z) \\
&= \sum_z (z-\mu)^2*p_z(z)
\end{align}
$
- *Method of moments:* In this simpler form we just need to subtract the $2^{nd}$ moment by the squared expectation.
$ \mathrm{Var}(X)=\mathbb E[X^2]- (\mathbb E[X])^2 $ ^559043
Derivation for method of moments:
$
\begin{align}
\mathrm{Var}(X) &= \mathbb E \big[(X-\mu)^2 \big] \tag{1}\\[4pt]
&= \mathbb E \big[(X^2-2\mu X + \mu^2) \big] \tag{2}\\[4pt]
&= \mathbb E[X^2]- 2\mu *\mathbb E[X] + \mathbb E[\mu]^2 \tag{3}\\[4pt]
&= \mathbb E[X^2]- 2\mathbb E[X]^2 + \mathbb E[X]^2 \tag{4}\\[4pt]
&= \mathbb E[X^2]- \mathbb E[X]^2 \tag{5}
\end{align}
$
where:
- (2) Expand the quadratic term.
- (3) Separate the expectations by linearity of expectations.
- (4) $\mu = \mathbb E[X]$.