RANDOM VARIABLES

Given a random experiment we are interested in a specific

quantity X thatdepends on the outcome ofthe experiment
he refer to as
X RANDOM VARIABLE. A RONDON VARIABL

isa
defined on the probability space (5,5,p)
x

function
#

witable X:SIR (or(RP).

EXAMPLI

⑨ Toss (coins. X: number

=

of heads
Number Head's
of
X 5
=
= IR
S HA 12
1 1r2
+

1T
31
&
- 1
16+ r
#
TH 01

TT
-

i r> 0
+

>0

② whether
Interest a
lightbulb's lifetime exceeds2 years.
lifetime
S
0 2
years
X: =

I lifetime 2
years
*
j
I o S (R>0
=

2
years X:(R >0 ·/R
-1
⑧
x1(9)(2,+x(t)
the INDICATOR FUNCTION 11g(x) is defined for
AC5as 1 xt A

\
aug
a(x(x) =

⑧ x *A
A
random variable X is when the set
DiscetE of possible
values of X finite
is or countable, i.e.

X(S) 5xy,22,
=

-, X23 or X(s) 3x3,x2.

=
-
3
A random variable is continuous ifthe it ofpossible
of
values X
isinfinite and not countable, e.g. x(5):IR
MIXED RANDOM VALLABIS

In the previous exampleswe have

⑨ X(s) 50,1,23 =
② x (s) 50,13 =
both discrete r.v.5

③ S: All possible financial marketscenarios

x =

pace of a specific stock x

(S)=1R>0 continuousr.V.

We are interested in of *
the distribution e.g. in answering 11

questionssuch "what is P(X32)?"or "what isP(X-51,23p

How
as

does P relate to X?
For a discrete random valiable
DF
1=
X on (5,2,P):
P(X x) P(5w-Ss.t.X(u) x3)a X(5)
=
= = = +

EXAMPL

In S =
[104, 15, TH, 53 and X. =theaols x (S) 0, 1,23
=

P(X 0) P(2++3) Yk
= = =

Intuitive.
Necessary definition as

P(X 1) P(915,TH3) 1/2

=
= =

we
originally defined P as
a

P(X 2) P(51513) 14
= =
=

function on

Thisconcept
generalizes to collimousriv.
We define the PROBABLY MASS FUNCTION (PMF)

py(x) p(x) of
= as:
X
pi X(5) > [0,19
⑧

DIS(8t +E x
,sp(X x) =

In for
example p111 =y2,p10) (,p(2)p
=

P(x)

1/4
·

12'x
For cotimousr.v.swe have P(X=x) Px - X(5).
We define
the PROBABCut Dinseth Function (PDF)

of a continuous rv. X as the

of
function such that

P((x(a,b)) sPf(x)0x
=

Va,b + X(5)

+sP(x-
(a,b))

>x

The CUMULA
DonniTY Function (CDF) F of
a continuous or discrete Mr. X is defined:
Fx(x) P(X-X)
=

Vx X(S)
-

or valve (realisation
 F(X) I
W
8.

3/4 -
⑧ 0

In the colf:"2-
11 0

I I I
X
I 2

For continuous the

r.v.5 colf lookslike

xit--------- x
A colf must
satisfy the
following properhes:
ii) F(x) hor-decreating funcion of
X

(ii) lim F(x) $ =

XN

Gil line F(X) 0

=

x--

NOTATION We demate with

rv.s capital letterse.g. X,4,E
andthe numerical values they can take by lowercase letters

e.g. x,y,z. When using pruf, polyor calf, we can

explicitly
highlight that
it refers to a certain r.v. X
by writing
Px(x), fx(x), Fy(x) rather than p(x), f(x), f(x).
The EXPECTED VALUE E(x] of a rr. X isdefined

E[X] =
= 5 xp(x) 5 xP(X x)
=
= Oliscrete
X

x X(5)
+
xX(S) +

*[x] is the
weighted average ofthe possible valuesofX,
its
Where the
weightof each value is
x
probability pxX).
+

E[x] =

Sx4xxidXfor X cautious

=[x] represents the boncentre of distribution.

the

For a riv. Var(x) is

the XARANDS
X debued:
Var(x) -[(x-5[x9)2)
=
=

5(x2)-e"[x]
The variance is a measure ofdistribution dispersion.
POPULAR FAMIUE5 r.V.
OF

DISCRETE CONTINUOUS

Uniform Uniform
Bernoulli Exponential Each destribution has
Binomial normal its own
prf/polf
Geauebric ↓-Student and colf
Poisson x2
I