Univariate Random Variables

W E R

Consider the experiment of tossing 3 coins

A HAH HTH TTT

Consider
the function X which counts the number of heads obtained

X R R
X w number of heads in

X AHH Wi 3
X HTT
etc X is basically a random variable

Definition A random variable is a function X A R

Definition i If S is a set and Y ES we denote the

indicator function of Y as

I 5 k
Iy w
to
if we 4
otherwise
 Indicator Random Variable

Definition Lef R be the sample space and A E R

be an eventWe define the indicator
random variable of A as

Ia P R

Iain
to Ine
I
This is also called a Bernoulli random variable with parameter
p PLA

Probability distribution of a random variable

X is a random variable
B E R
We are interested in calculating Pr X C B

Act EB P wer X w c B

Eg Tossing a con 3 times X number of heads

f X 1 P wer Mw I
 I
P HTT THT ITH lD 38

Definition If X is a random variable we define the

probability distribution or law of X as

PhxEB BER

Example Consider the of tossing 3 coins numberofheads

experiment X

f X cB P WER X w c B

The law of total probability gives us

Pdx EB P X EB n f a

IE PLEX a Iola

Definition Let X be a random variable The cumulative

distribution function of X is the function
E R 0,1
defined by
Else Pr Xen
IP Xue the ER
 Theorem The CDF characterizes the distribution of a random

variable If you know the calf then know

you
the distribution and use versa

Example
X number of heads in 3 fair corn tosses

Fx a PLEX ex
IIa Ge t
III be IIa be ftp.agbe

Theorem If Fx is a calf then Fx is

1 Increasing ie
Ney Fx a E Fxly Vu y c IR

2 Rightcontinuous ie fxCx the C R

ylz fxCy
3 I fxCn O and L fix 1

Recovering some probabilities from the CDF

Theorem Let X be any random variable Then

Cis for any acb IP a axe to Ecb fx a

ii if celR then IPEX c file ECE

 I

ii implies that the of the jump of the CDF at

size
the point c is equal to the probability that X i

Fx is continuous at cell lPC X c O

A point c such that IP X c 0 is called

an atom of Fx

Types of Random Variables

Definition A random variable X is discrete if there

exists a countable set Ni Nz such
that
Ep xn3 1

We write Ronge X Ni Nz

for a discrete random variable X we define he

probability mass function PMF of to be the
function IR 0,1 such that
p
pin PC X se for NER
 A discrete rv is one that has countably many atoms
whose probabilities add
up to
1

Definition A random variable is said to be continuous

if Fx is a continuous function We
say
that X is absolutely continuous if
FX is an absolutely continuous function
in the that there exists and integrable
sense
function f such that i

Ecb fx a
fabfxlnldnVa.be R
The function fx above is called a probability density
function PDF
for continuous random variables IP X c 0 However
this does not imply that X never takes the value C

Normal Gaussian distribution

A random variable
X is normally distributed if X
is continuous with density
absolutely

1 ICTY
Aca FE e KEIR

L
 r
y n N Lte 62

If we have U O and 6 then X is 1

said to have the standard normal distribution Its

density is denoted by and its CDF by OI

th
0cal e NER

Ela f.IO dt seek

Note OI is the Cdf of the STANDARD NORMAL ONLY

not just any normal distribution

Uniform distribution

let AER have positive total

length We say
X is uniformly distributed on A if X is
absolutely continuous with density

Ata t.tw ofIItGe

We write X n
U Ca b for A a b
 the density can
simply be written as

frfr Ica b be
b a

Gamma distribution

Definition We define the gamma function P an IR by

Ma foot e tdt se 0

Properties D Mi I
2 Platt Nra N O
1
3 f n n l n c IN
4 PFE Tn
Definition A randomvariable X has a
gamma
distribution with shape parameter
2 0 and scale parameter 1 0
if X is absolutely continuous with
density
en
lira
Aca nca
e
Ico a N ne IR
 We write Xn Gamma x l

If 2 1 then Gamma 1,1 Exp t

If l I and a Z the Gamma EE X

Chi squared distribution with n degrees of freedom

If a _n is an integer then Gamma nit is called

the Erlang distribution and is the distribution of the
arrival time of the nth event when events arrive according
to a Poisson
process

Beta Distribution

Definition
B 0 a x o.o IR is defined by

B mm
fit Li tf dt

Properties i B Min Bhim

2 B Min Cm Th
r ntm

Definition
X has a Beta distribution with parameters
 p
Min 0 if

frfr Nm Ct 2e
Ico Ca
Fy
A random variable with Beta 1,1
distribution is

actually the uniform distribution between Omd 2 Uco17

Other Random Variables

Consider
X Exp I
Y XI x 53

Find the CDF of Y and state whether 4 is discrete

Solution

Y X if x 5
0 if NES

Fy y lPC
Yey
we use the LOTP
IP Yey n Ex 53 t PC yey3n sees
 y y
when zess we know that X Y so we can write
IP xey3n x 53 t PC
yey3n sees
when N S 4 0 so we can substitute into the 2ndterm
IP xey3n x 53 t P n sees
oey
P sixty Phooey n sees

H scxey3 0 yes
Fxly fils y
S

p oey3n nests
pquesz by
o

L c

i We have Fy y O y co

lie
1 e
i
T
e s
e O s
S
y

The CDF has one atom at y O for it to be

discrete it must have a countable number of
atoms whose probabilities add up to 1

We can conclude that 4 is not discrete

Since the CDF is not continuous we

may
conclude that Y is not absolutely continuous