IE 265

• Random Variables – Part 3

• Characteristics of distributions
• Expected value (mean)
• Variance and standard deviation
• Chebyshev’s inequality
 2

Characteristics of distributions
Expected value or mean
• Expected value or mean is a measure of central
tendency of a random variable.
• It is denoted by 𝐸(𝑋) or 𝜇 and found as
E ( X )     xp( x), if X is discrete
x

  xf ( x)dx,

if X is continuous

• 𝐸(𝑋) is the weighted (by probability) average of possible

values 𝑋 can take on.
• 𝐸(𝑋) is the “population” mean as opposed to “sample”
mean.
 3

Characteristics of distributions
• Ex: Rolling a die 3 5
.
sample size4
average close
X: value on the upper face
to 3 5
What is the expected value of X? ,

• Ex: Tossing an unfair coin with P(H) = 1/4 and

P(T) = 3/4. If it is H we win $8, and if it is T we lose $1.
X: value we win
What is the mean of X?

E
E(X) 8
44 44 3/4 74
=
=
X 8=
.

2 -

P(x)
1
3/4 X =
-

Ow
20
 4

Characteristics of distributions
• Ex: Weekly kerosene demand has the pdf

E
 x, 0  x 1

f ( x)  1 / 2, 1  x  2
0, otherwise

What is the expected weekly demand?
=
13/12
• Ex: Find the mean of random variable X having pdf L
8

e  x , x0
du d= f ( x)  
-ebx 0, x0

=
V=
A

E(X) =

S = -X
 5

Characteristics of distributions
Variance and standard deviation
•
10
Variance is a measure of dispersion (spread) of a random
- =
(0 - a -

variable
• It is denoted by 𝑉𝑎𝑟(𝑋) or 𝜎 2 and found as
=
Ha
Var ( X )   2  E{[ X  E ( X )]2 }
  ( x   ) 2 p( x), if X is discrete
x

  ( x   ) 2 f ( x)dx, if X is continuous


• 𝑉𝑎𝑟(𝑋) is the expected squared deviation of 𝑋 from its

expectation.
L units are
(r v)2 .
 6

Characteristics of distributions is
Variance and standard deviation unit some
- wrv
• Standard deviation is 𝑆𝑡𝑑 𝑋 = 𝜎 = 𝑉𝑎𝑟(𝑋).
• It can be thought of as the magnitude of typical deviation
between an observed outcome and its expectation.

• 𝑉𝑎𝑟 𝑋 = 0 iff 𝑋 takes on a single value, i.e. 𝑋 takes on a

constant value with probability 1 and has no variability.

bu 0 El = 0 mu dand

coefficient of variation is (V = 0
-

N
dimensionless
"
 7

Characteristics of distributions
• Ex: Tossing a fair coin with P(H) = P(T) = 1/2
0
Game A: if H win $1, if T lose $1 +12
E(X) = 2 .
-
1 .

+z =

Game B: if H win $100, if T lose $100 E(x) 1/2 100 = , -

12 1000
,

What is the expected payoff of games A and B?

Which game has more variability in payoff? -

V(X) =

• Ex: Find the variance of random variable X having pdf

4 x 3 , 0  x 1
f ( x)  
0, otherwise

• Ex: Find the standard deviation of weekly kerosene

demand.
 8

Chebyshev’s inequality
Original name: Tchebysheff
• Theorem: Let 𝑋 be a (discrete or continuous) random
variable with mean 𝜇 and variance 𝜎 2 . Then, for any
positive 𝑘 > 1
1 1
P (| X   | k )  2 or P (| X   | k )  1  2
k k
• This theorem provides a lower 𝑓(𝑥)
bound on the probability that
𝑋 lies within the interval
(  k ,   k ).

• It does this without any

𝑥
information about the 0 
distribution of 𝑋.   k   k
 9

Chebyshev’s inequality
• Ex: The # of customers per day at a certain sales counter,
denoted by Y, has been observed for a long time. Y is
found to have a mean of  = 20 customers and a standard
deviation of  = 2 customers.
a) What can be said about the probability that Y will be between 16
and 24 tomorrow?
b) Find the shortest interval that is certain to contain at least 90%
of the customer levels.
c) How does your answer to part a change if the standard
deviation is  = 1?
 10

Chebyshev’s inequality
• Ex: Daily cost for use of a certain tool has a mean of  = $15
and a variance of 2 = 36. How often will this cost exceed
$25?