DerivedDistro

Derived Distribution

Compose one or more simple random variables to get more complicated r.v. (This is how we go from simple model to complicated model) ($f(X)$)

Discrete

case

Obvious formula

$$
P_Y(y) = P(Y=y) = \sum_{\text{all x if g(x)=y}} P_X(x)
$$

$$
P_Y(4) = P_X(5)+P_X(4) = 0.4 + 0.3 = 0.7
$$

Linear Example

img

The probability doesn’t change at all, it is just a remapping

Continuous

For linear functions, it is just a similar story

sim

$$
Y = aX+b
$$

$$
f_Y(y) = \frac{1}{|a|}f_X(\frac{y-b}{a})
$$

Deduction

deduction

Application

we can use this formula to see that linear function applied on a normal distribution still gives a normal distribution

$$
if X \sim N(\mu , \sigma^2 ) , Y = aX+b (a\ne 0)
$$

$$
then , Y \sim N(a\mu +b , \sigma^2a^2)
$$

General/Systematic way to find any function applied on a known X

  1. Find $F_Y(y) = P(g(X) \le y)$

  2. Take derivative $f_Y(y) = \frac{dF_Y(y)}{dy}$