You are on page 1of 28

Chapter 2: Random Variable

and Probability Distributions

Yang Zhenlin

zlyang@smu.edu.sg
http://www.mysmu.edu/faculty/zlyang/

Chapter Contents

Chapter 2

Random Variable (r.v.)

Discrete r.v.
Continuous r.v.

Distribution of a Random Variable

Probability mass function


Probability density function

Expectation of a Random Variable


Variance and Standard Deviation
Moment Generating Function

STAT151, Term I 12-13

Zhenlin Yang, SMU

Random Variable

Chapter 2

In real-world problems, we are often faced with one or more


quantities that do not have fixed values.
The values of such quantities depend on random actions and
they change from one experiment to another,
Number of babies born in a certain hospital
Number of traffic accidents occurred on a road per month
The amount of rainfall in Singapore per year
The starting price of a stock each day, etc.

In probability, quantities introduced in these diverse examples


are called Random Variables.
Study the properties of random variables allows us to have a
better understanding on the real-world phenomena, and to
control or to predict their behavior, .
STAT151, Term I 12-13

Zhenlin Yang, SMU

Random Variable

Chapter 2

Example 2.1 If in the experiment of rolling two fair dice, X is the


sum, then X is a variable. Since the value of X changes from one
experiment to another, X is also a "random variable". The possible
values of X and the corresponding probabilities are summarized as
follows:

where, for example,


p2 = P(X=2) = P{(1,1)} = 1/36,
p3 = P(X=3) = P{(1,2), (2,1)} = 2/36,
p7 = P(X=7) = P{(1,6), (6,1), (2,5), (5,2), (3,4), (4,3)} = 6/36
STAT151, Term I 12-13

Zhenlin Yang, SMU

Random Variable

Chapter 2

In Example 2.1, numerical values of X depend on the


outcomes of the experiment, e.g.,

if the outcome e = (2,3), then X = 2+3 = 5,


if e = (5,6), then X = 5+6 = 11, etc.

Thus, X is a real-valued function.


On the other hand, if, for example, X = 3, then it must be that
e = (1,2) or (2,1). Therefore, the inverse image of the function
X is an event!
Definition 2.1. Let S be the sample space of an experiment. A
real-valued function X defined on S and taking values in R
(the set of real numbers) is called a random variable (r.v.) if
for each interval I R, the inverse image {e: X(e) I } is an
event.
STAT151, Term I 12-13

Zhenlin Yang, SMU

Chapter 2

Random Variable

Example 2.2 Suppose that we toss a coin having a probability p of coming up


heads, until the first head appears. Let N denote the number of flips required,
then assuming the outcome of successive flips are independent, N is a random
variable taking values 1, 2, 3, , with respective probabilities
P{N =1} = P(H) = p,
P{N =2} = P(TH) = (1p)p,
P{N =3} = P(TTH) = (1p)2p,

n 1
P{N n} P (TT

T
H
)

(
1

p
)
p

n 1

All probabilities add up to 1: p n 1 (1 p)

STAT151, Term I 12-13

n 1

1
1 (1 p )
Zhenlin Yang, SMU

Distribution of a Random Variable

Chapter 2

Definition 2.2 (Cumulative Distribution Function) The


cumulative distribution function (CDF) of a random variable X
is defined by,
F(x) = P{X x},
for any real x.
F(x) cumulates all of the probabilities of the values of X up to
and including x.
A function F(x) is a CDF if and only if it satisfies:
(a)
limx F(x) = 0 and limx F(x) = 1
(b)

limh0 F(x+h)= F(x), h > 0

(right continuous)

(c)

a < b implies F(a) F(b)

(nondecreasing).

STAT151, Term I 12-13

Zhenlin Yang, SMU

Distribution of a Random Variable

Chapter 2

Example 2.3. A four-sided die has a different number, 1, 2, 3, or


4, affixed to each side. On each roll, each of the four numbers is
equally likely to occur. A game consists of rolling the die twice.
If X represents the maximum of the two rolls, then X is a
random variable. Find and sketch the CDF of X
Solution: Possible values for X are 1, 2, 3, and 4, and the corresponding
probabilities are:
P(X = 1) = P{(1,1)} = 1/16
P(X = 2) = P{(1,2), (2,1), (2,2)} = 3/16
P(X = 3) = P{(1,3), (2,3), (3, 3), (3,2), (3,1)} = 5/16
P(X = 4) = P{(1,4), (2,4), (3,4), (4,4), (4,3), (4,2), (4,1)} = 7/16.
Thus, the CDF of X is,
x
F(x)

1/16

4/16

9/16

STAT151, Term I 12-13

Zhenlin Yang, SMU

Distribution of a Random Variable

Chapter 2

Example 2.4. A function is specified as follows.


0
x0
0 x 1
x 4
12
1 x 2
F(x) =
x 12 1 2 2 x 3
1
x 3.

(a) Verify that it is a CDF.


(b) Plot F(x) and calculate: P(X < 2),
P(X = 2), and
P(1 X < 3).

Solution: (a) ,
F(x)
1

(b) : P(X < 2) = F(2) = 1/2.


P(X = 2) = F(2) F(2)
= (2/12+1/2)1/2 = 1/6.

3/4
2/3
1/2

P(1 X < 3) = P(X < 3) P(X <


1)
= F(3) F(1)
= (3/12+1/2) 1/4 = 1/2.
STAT151, Term I 12-13

1/4
0
9

Zhenlin Yang, SMU

Distribution of a Random Variable

Chapter 2

Note. Random variables can be classified as discrete, continuous,


or mixed. The function given in Example 2.4 is a mixed CDF
which has jumps at certain points. We will concentrate on pure
discrete and pure continuous types of random variables.
Definition 2.3 (Discrete R.V. and Probability Mass Function)
If the set of all possible values of a random variable, X, is a
countable set, x1, x2, . . . , xn, or x1, x2, . . ., then X is called a
discrete random variable. The function
p(x) = P(X = x),

x = x1, x2, . . .

that assigns the probability to each possible value of X will be


called probability mass function (pmf).
So, pmf tells how the probabilities are distributed across the values of r.v. X.
STAT151, Term I 12-13

10

Zhenlin Yang, SMU

Distribution of a Random Variable

Chapter 2

A function p(x) is a pmf if and only if


(i) p(x) 0 for all x,

and
(ii) p ( x) 1.
all x

Example 2.5. Is each of the following functions a pmf? If it is,


find the constant k.
(a)
p(x) = k(1/2)x , x = 0, 1, 2, . . .
(b)

p(x) = k [(1/2)x 1/2], x = 2, 1, 0, 1, 2.

Solution: (a) p(x) > 0 for all x = 0, 1, 2, . . . ; and


2
p
(
x
)

k
(
1

1
/
2

(
1
/
2
)
) 2k , k 1 / 2

all x

(b) Since p(0) = k/2, and p(2) = k/4, the two values of the function have
opposite signs no matter what value k takes. Therefore, it is impossible for
p(x) to be always nonnegative, and hence p(x) cannot be a pmf.
STAT151, Term I 12-13

11

Zhenlin Yang, SMU

Distribution of a Random Variable

Chapter 2

The r.v's introduced above take countable number of values.


However, many social phenomena can only be described by
variables with uncountable values, e.g., arrival time of a train,
lifetime of a transistor, etc.
Definition 2.4 (Continuous R.V. and Probability Density
Function) A r.v. X is said to be a continuous random variable if
there exists a nonnegative function f, defined for all real x (
, ), such that for any set B of real numbers
P{X B} =

The function f is called the probability density function (pdf) of


the r.v. X.
So, pdf f(x) tells how likely values of r.v. X occurs at around x.
STAT151, Term I 12-13

12

Zhenlin Yang, SMU

Distribution of a Random Variable

STAT151, Term I 12-13

13

Chapter 2

Zhenlin Yang, SMU

Distribution of a Random Variable

Chapter 2

Example 2.6 A CDF has the following expression


F(x) =
Sketch the graph of F(x) and find the pdf f(x).
Solution:
The graph of F(x) is shown below.

Clearly, F(x) is not differentiable at x


= 0, 0.5 and 1.5. Take the right
derivatives, i.e., f(0) = 1, f(0.5) = 0.5,
and f(1.5) = 0, we then have

F(x)
1

0,
1,
f ( x)
0.5,

0,

.5
0

.5

STAT151, Term I 12-13

1.5

x
14

x0
0 x 0.5
0.5 x 1.5
1.5 x
Zhenlin Yang, SMU

Distribution of a Random Variable

Chapter 2

A function f(x) is a pdf if and only if it satisfies the properties


(i) f(x) 0 for all x

and

(ii)

f ( x)dx 1

Example 2.7 A machine produces copper wire, and occasionally


there is a flaw at some point along the wire. The length of the
wire (in meters) produced between successive flaws is a
continuous r.v. X with pdf of the form:
f(x) = c(1+x)3, x > 0,
where c is a constant. Find c, and CDF.

STAT151, Term I 12-13

15

Zhenlin Yang, SMU

Expectation of a Random Variable

Chapter 2

Definition 2.5. (Expectation) The expected value of a discrete r.v.


X with pmf p(x) is defined by

= E(X) =

x p( x),
all x

i.e., a weighted average of all possible values of X. It is also called


mean of the population represented by X.
Expected value of a discrete r.v. is just the weighted average of the
possible values of X, weighted by their chances to occur.
The expected value of a continuous r.v. with pdf f(x) is defined by

= E(X) =

x f ( x)dx,

if the integral is absolutely convergent. Otherwise we say that


E(X) Termdoes
STAT151,
I 12-13 not exist.
Zhenlin Yang, SMU
16

Expectation of a Random Variable

Chapter 2

In Example 2.1, X is the sum of the two numbers shown on the


two dice. The expected sum is:
1
2
3
1
E(X)
=x p( x) (2) (3) (4) (12) 7

36
36
36
36
all x

Expectation is an extremely important concept in summarizing


characteristics of distributions. It gives a measure of central
tendency. It has following simple but important property:
If a and b are constants, then
E[aX + b] = a E[X] + b
STAT151, Term I 12-13

17

Zhenlin Yang, SMU

Chapter 2

Function of a Random Variable


The following results are important:
If X is a r.v., then a function of it, u(X), is also a r.v. ,
E[u(X) ] = u ( x) p( x) , if X is a discrete r.v. with pmf p(x),
E[u(X) ] =

all x

u ( x) f ( x)dx , if X is a continuous r.v. with pdf f(x).

Example 2.8 The r.v. X has pmf p(x) = 1/3, for x = 1, 0, 1. Find
E(X) and E(X2).
Solution:

E ( X ) xp( x) (1) 13 (0) 13 (1) 13 0


all x

E ( X ) x p ( x) (1)
2

all x

(0) (1)

2 1
3

2 1
3

2 1
3

The above result says that the expectation of the r.v., u(X), can
simply be found through the distribution of the original r.v. X !
STAT151, Term I 12-13

18

Zhenlin Yang, SMU

Chapter 2

Variance of a Random Variable

Another important quantity of interest is the variance of a r.v. X,


denoted by Var(X), which is defined by
Var(X) = E[(X )2],
where = E(X). A related quantity is the standard deviation,

(X) = {Var(X)}1/2.
Variance and standard deviation (sd) measure how much the
values of X vary around its mean .
As (X )2 is a function of X, i.e., u(X) = (X )2, from the results
given in the last slide, we have
Var(X) =
Var(X) =
STAT151, Term I 12-13

, if X is discrete with pmf p(x),


, if X is continuous with pdf f(x).
19

Zhenlin Yang, SMU

Variance of a Random Variable

STAT151, Term I 12-13

20

Chapter 2

Zhenlin Yang, SMU

Variance of a Random Variable

Chapter 2

We note in Example 2.9 Var(X) and Var(Y) are the same. Why?
If X is a r.v. with mean , and a and b are arbitrary constants, then
E(aX + b) = aE(X) + b
Var(X) = E(X2) 2

To show third result,


Var(aX + b)
= E[(aX + b E[aX + b])2]
= E[(aX + b aE[X ] b)2]
= E[(aX aE[X ])2]
= E[a 2(X E[X ])2]
= a 2E[ (X E[X ])2]
= a2Var(X)

Var(aX + b) = a2Var(X)

(aX + b) = |a|(X).
To show the second result,
Var(X) = E[(X )2]
= E(X2 2X+ 2)
= E(X2) 2 E(X)+ 2
= E(X2) 2
STAT151, Term I 12-13

21

Zhenlin Yang, SMU

Variance of a Random Variable

Chapter 2

Example 2.10. The monthly sales at a computer store have a


mean of $25,000 and a standard deviation of $4,000. Profits are
30% of the sales less fixed costs of $6,000. Find the mean and
standard deviation of the monthly profit.
Solution: Let X = Sales. E(X) = 25,000, and Var(X) = 4,0002
Let Y = Profit. Then, Y = .30(X) 6,000.
E(Profit) = E(Y)
= .30 E(X) 6,000
= (.30)(25,000) 6,000 = 1,500.
Var(Profit) = Var(Y)
= Var[(.30)(X) 6,000]
= (.30)2Var(X) = 1,440,000

(Profit) = (Y) = 0.3(X) = 1,200.


STAT151, Term I 12-13

22

Zhenlin Yang, SMU

Moment Generating Function

Chapter 2

The mean, variance and standard deviation are important


characteristics of a distribution.
For some distributions, it is rather difficult to compute these
quantities directly.
A special function defined below can help. More importantly,
the uniqueness property of this function often help to find the
distribution of some useful functions of r.v.s.
Definition 2.12 (Moment-Generating Function) Let X be a
random variable, discrete or continuous. If there exists a positive
number h such that
M(t) = E(etX)
exists and is finite for h < t < h, then M(t) is called the MomentGenerating Function (MGF) of X.
STAT151, Term I 12-13

23

Zhenlin Yang, SMU

Chapter 2

Moment Generating Function


Why is it called the Moment Generating Function?
because it generates moments.
What are the moments?

the term moment is from mechanics, representing the product


of a distance and its weight.
So, xi p(xi) is a moment, and xi p(xi) or E(X) is the moment of
the system.
In statistics, E(X), E(X2), E(X3), , are the 1st, 2nd , 3rd, ,
moments about the origin;
E(X), E[(X)2], E[(X)3], etc, are the 1st, 2nd , 3rd central
moments, or moments about the mean.
See p67 of text for details.
STAT151, Term I 12-13

24

Zhenlin Yang, SMU

Chapter 2

Moment Generating Function


Property of MGF:

M.G.F., if it exists, completely determines the distribution


function. In other words, if two random variables, assuming the
same set of vaules, have the same M.G.F., they must have the
same distribution function.
E(Xr) = M(r)(0), the rth derivative of M(t) evaluated at 0.
We use a discrete r.v.s to demonstrate these properties: if X and Y
have possible values {v1, v2, }, and pmfs p(x) and q(x), then,

M X (t ) E (etX ) et v1 p (v1 ) et v 2 p (v2 ) ,


M Y (t ) E (etY ) et v1 q(v1 ) et v 2 q (v2 ) .
Thus, if MX(t) = MY(t), it must be that
p(vi) = q(vi), i = 1, 2,
STAT151, Term I 12-13

25

Zhenlin Yang, SMU

Moment Generating Function

Chapter 2

To understand the property E(Xr) = M(r)(0), we have

M (t )

d
dt

t x1

p ( x1 ) et x 2 p( x2 ) et x3 p( x3 )

et x1 x1 p( x1 ) et x 2 x2 p( x2 ) et x3 x3 p( x3 )
Thus, M (0) x1 p ( x1 ) x2 p ( x2 ) x3 p ( x3 ) E ( X )

M (t )

d2
dt 2

t x1

p ( x1 ) et x 2 p ( x2 ) et x3 p ( x3 )

et x1 x12 p ( x1 ) et x 2 x22 p ( x2 ) et x3 x32 p ( x3 )


2
2
2
2

M
(
0
)

x
p
(
x
)

x
p
(
x
)

x
p
(
x
)

E
(
X
)
Thus,
1
1
2
2
3
3

In general,

(r )

(t ) i 1 e x p ( xi ),

STAT151, Term I 12-13

t xi

r
i

M
26

(r )

(0) i 1 xir p ( xi ) E ( X r )

Zhenlin Yang, SMU

Moment Generating Function

Chapter 2

Example 2.11. If X has the MGF,


3
2t 2
3t 1
M (t ) e e e
6
6
6
t

then, as the coefficients of the e terms are probabilities, the


probabilities must be 3/6, 2/6, and 1/6; the values beside t are the
values of X, which are 1, 2, and 3. The pmf of X is
x

p(x)

3/6

2/6

1/6

Example 2.12. Suppose the MGF of X is


et 2
M (t )
, t ln 2.
t
1 e 2
Find the distribution of X.
STAT151, Term I 12-13

27

Zhenlin Yang, SMU

Moment Generating Function

Chapter 2

Solution: Until we expand M(t), we cannot detect the


tx
coefficients of e i .

1
Recall:
1 z z 2 z 3 ,
1 z

1 z 1

we have,
That is, P(X = x) = (1/2)x,
for positive integer x; the
pmf of X is thus,

e 2
M (t )
1 et 2

e t e t 2 e 3t
1 2 3

2
2
2

et

1
1
e 2t
2
2

et
STAT151, Term I 12-13

1

2

1
p( x) , x 1,2,3, .
2

e3t

28

Zhenlin Yang, SMU

You might also like