# Random variable sets

Jump to: navigation, search
 Please improve this article by expanding it. A short introduction is required... what is a random variable set? There are only formula here...
 This article does not cite any references or sources. Please help improve this article by adding citations to reliable sources.

A random variable set is a vector of Random variables $\boldsymbol X = {X_{1},..., X_{n}}$, where $\boldsymbol X_{1},..., X_{n}$ do not necessarely have the same distribution

#### Two random variables

Two events X and Y are independent if:

$P(X\cap Y) = P(X) \cdot P(Y)$

For continuous probabilities, the previous equation is expressed as:

$f(x,y)=f_{X}(x) \cdot f_{Y}(y)$

where $f_{X}$ and $f_{Y}$ are the marginal density functions, defined as

$f_{X}(x) = \int_{-\infty}^{+\infty} f(x,y) dy$
$f_{Y}(y) = \int_{-\infty}^{+\infty} f(x,y) dx$

Let Y have a probability greater than zero. The probability of the event X assuming that Y occurred is (X and Y may not be independent):

$P(X| Y) = \frac{P(X\cap Y)}{P(Y)}$
$P(X| Y) = \frac{P(Y | X)\cdot P(X)}{P(Y)}$

If X and Y are continuous:

$f_{X|Y}(x, y)= \frac{f(x,y)}{f_{Y}(y)}$

#### Several random variables

Let $\boldsymbol X= \left( X_{1}, ..., X_{n} \right)$ be a vector of $n$ random variables. The mean and the standard deviation of the $i^{th}$ term are:

$\mu_{X_{i}} = E(X_{i}) = \int_{-\infty}^{+\infty} ... \int_{-\infty}^{+\infty} x_{i} f_{\boldsymbol X}(\boldsymbol x) dx_{1} ...dx_{n}$
$\sigma_{X_{i}}^{2} = E\left( (X_{i}- \mu_{X_{i}}) ^{2}\right) = \int_{-\infty}^{+\infty} ... \int_{-\infty}^{+\infty} \left( x_{i} - \mu_{X_{i}} \right)^{2} f_{\boldsymbol X}(\boldsymbol x) dx_{1} ...dx_{n}$

The covariance of the random variable $X_{i}$ and $X_{j}$ is defined as:

$Cov(X_{i},X_{j}) = E\left( (X_{i} - \mu_{X_{i}})(X_{j} - \mu_{X_{j}})\right)$

It can be shown that:

$Cov(X_{i},X_{j}) = E\left( X_{i} X_{j} \right) - E\left( X_{i} \right) E\left( X_{j} \right)$

The covariance measures the relationship between two random variable. If $Cov(X_{i},X_{j})>0$, $X_{i}$ is likely to be greater (resp. lesser) than its mean value if $X_{j}$ is greater (resp. lesser) than its mean value. If $Cov(X_{i},X_{j})<0$, $X_{i}$ is likely to be lesser (resp. greater) than its mean value if $X_{j}$ is greater (resp. lesser) than its mean value. Independent random variables are uncorrelated, whereas uncorrelated random variables are not necessary independent.

The covariance matrix $\boldsymbol C$ is defined as:

$\boldsymbol C = \left[ Cov(X_{i},X_{j})\right] _{i,j = 1...n}$

The correlation $\rho$ is expressed as:

$\rho = Corr(X_{i},X_{j}) = \frac{Cov(X_{i},X_{j})}{\sqrt[]{Cov(X_{i},X_{i})\cdot Cov(X_{j},X_{j})}} = \frac{Cov(X_{i},X_{j})}{\sigma_{X_{i}} \cdot \sigma_{X_{j}} }$

The value of the correlation is in the range [-1, 1]. It is a scaled measure of the relationship among random variables

The probability density function of $n$-dimensional Gaussian distribution can be determined using the covariance matrix and a vector containing the mean of each random variable $\boldsymbol\mu_{\boldsymbol X }$ (the $i^{th}$ term being the mean value of $X_{i}$):

$f(\boldsymbol x ) = \frac{1}{(2\pi)^{\frac{n}{2} }det(\boldsymbol C )^{\frac{1}{2}}} \cdot exp\left(-\frac{1}{2} (\boldsymbol x - \boldsymbol\mu_{\boldsymbol X } )^{T} \boldsymbol C ^{-1} (\boldsymbol x - \boldsymbol\mu_{\boldsymbol X } ) \right)$

Random variables