Statistics

CVEN2002/2702

Week 5

This lecture

5. Special random variables

Additional reading: Sections 1.6 + pp. 28-29, 32-34 in the textbook

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

2/1

5. Special random variables

5. Special random variables

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

3/1

5. Special random variables

5.1 Introduction

Introduction
In practice, we often run the same kind of (random) experiments, from
which we are often interested in the same kind of results
it turns out that certain types of random variables come up over
and over again in applications
In this chapter, we will study a variety of those special random
variables
You can also go to
http://socr.stat.ucla.edu/htmls/SOCR_Distributions.html
and have a look at the numerous special distributions there

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

4/1

5. Special random variables

5.2 Binomial random variables

Example
Consider the following random experiments and random variables:
flip a coin 10 times. Let X = number of heads obtained
a worn machine tool produces defective parts 1% of the time . Let
X = number of defective parts in the next 25 parts produced
each sample of air has a 10% chance of containing a particular
molecule. Let X = the number of air samples that contain the
molecule in the next 18 samples analysed
of all bits transmitted through a digital channel, 15% are received
in error. Let X = the number of bits in error in the next five bits
transmitted
a multiple-choice test contains 10 questions, each with 4 choices,
and you guess at each question. Let X = the number of questions
answered correctly
similar experiments, similar random variables
a general framework that include these experiments as particular
cases would be very useful
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

5/1

5. Special random variables

5.2 Binomial random variables

The Binomial distribution

Assume:
the outcome of a random experiment can be classified as either a
Success or a Failure ( S = {Success, Failure})
we observe a Success with probability
n independent repetitions of this experiment are performed
Define X = number of Successes observed over the n repetitions. We
say that X is a binomial random variable with parameters n and :
X Bin(n, )
See that SX = {0, 1, 2, . . . , n} ( discrete r.v.) and the binomial
probability mass function is given by
! "
n x
p(x) =
for x SX
(1 )nx
x
# $
where xn is the number of different groups of x objects that can be
chosen from a set of n objects (Week 3)
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

6/1

5. Special random variables

5.2 Binomial random variables

The Binomial distribution

# $
Note: the coefficients xn = n!/(x!(n x)!) are called the binomial
coefficients, they are the coefficients arising in the famous Newtons
binomial expansion
n ! "
%
n k nk
n
(a + b) =
a b
k
k =0

These coefficients are often represented in Pascals triangle, named

after the French mathematician Blaise Pascal (1623-1662)

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

7/1

5. Special random variables

5.2 Binomial random variables

The Binomial distribution: pmf and cdf

1.0
0.6

0.8

1.0
0.8

pX(x)
0.4

pX(x)
0.2

0.2

0.2

0.2

0.2

0.4

p = 0.9

0.6

0.8
pX(x)
0.4

0.4

pX(x)

0.6

0.8
0.6

0.8
0.6
0.4

pX(x)

p = 0.8

1.0

p = 0.5

1.0

p = 0.2

1.0

p = 0.1

0.0

0.6
0.4

FX(x)

0.6
0.4

0.4

0.2

0.2

0.2

3
x

0.0

0.0

0.0

0.0

0.0
2

0.2

0.2

FX(x)

0.6
FX(x)

0.6
FX(x)

0.8

0.8

0.4

0.4

FX(x)

0.6

0.8

0.8

1.0

1.0

1.0

0.0

0.0

1.0

0.8

1.0

0.0

0.0

3
x

Binomial pmf and cdf, for n = 5 and = {0.1, 0.2, 0.5, 0.8, 0.9}
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

8/1

5. Special random variables

5.2 Binomial random variables

The Bernoulli distribution

Particular case: if n = 1

the Bernoulli distribution,

named after the Swiss scientist Jakob Bernoulli (1654-1705)

(see Slide 13 Week 4)
X Bern()
pmf:

1 if x = 0

if x = 1
p(x) =

0
otherwise

Note: if X Bin(n, ), we can represent it as

X =

n
%

Xi

i=1

where Xi s are n independent Bernoulli r.v. with parameters

each repetition of the experiment in the Binomial framework is
called a Bernoulli trial
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

9/1

5. Special random variables

5.2 Binomial random variables

Binomial distribution: properties

First note that
%

p(x) =

xSX

n ! "
%
n
x=0

x (1 )nx = ( + (1 ))n = 1

using the binomial expansion (expected result! Slide 10, Week 4)

Second, it is easy to see that if X1 Bin(n1 , ), X2 Bin(n2 , ) and X1
is independent of X2 , then
X1 + X2 Bin(n1 + n2 , )

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

10 / 1

5. Special random variables

5.2 Binomial random variables

Binomial distribution: expectation and variance

Recall the representation X =

)n

i=1 Xi ,

with Xi Bern()

We know (Slides 24 & 32 Week 4) that

E(Xi ) =
and
Var(Xi ) = (1 )
#) n
$ )n
)n
It follows E(X ) = E
i=1 Xi =
i=1 E(Xi ) =
i=1 = n and
+
* n
n
n
%
%
%
ind.
Var(Xi ) =
(1 ) = n(1 )
Xi =
Var(X ) = Var
i=1

i=1

i=1

Mean and variance of the binomial distribution

If X Bin(n, ),
= E(X ) = n
CVEN2002/2702 (Statistics)

and

2 = Var(X ) = n(1 )

Dr Justin Wishart

Session 2, 2012 - Week 5

11 / 1

5. Special random variables

5.2 Binomial random variables

Binomial distribution: examples

Example
It is known that disks produced by a certain company will be defective with
probability 0.01 independently of each other. The company sells the disk in
packages of 10 and offers a money-back guarantee if more than 1 of the
disks is defective. a) In the long-run, what proportion of packages is
returned? b) If someone buys three packages, what is the probability that
exactly one of them will be returned?
a) Let X be the number of defective disks in a package. Then, it is clear that
X Bin(10, 0.01)
Hence,
P(X > 1) = 1 P(X = 0) P(X = 1)
! "
! "
10
10
0.011 0.999 0.005
0.010 0.9910
=1
1
0
in the long-run, 0.5 percent of the packages will have to be replaced
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

12 / 1

5. Special random variables

5.2 Binomial random variables

Binomial distribution: examples

Example
It is known that disks produced by a certain company will be defective with
probability 0.01 independently of each other. The company sells the disk in
packages of 10 and offers a money-back guarantee if more than 1 of the
disks is defective. a) In the long-run, what proportion of packages is
returned? b) If someone buys three packages, what is the probability that
exactly one of them will be returned?
b) Let Y be the number of packages that the person will have to return. We
have
Y Bin(3, )
where is the probability that a package is returned, that is, contains more
than 1 defective disk. In a), we found that = 0.005
Thus, the probability that exactly one of the three packages will be returned is
! "
3
P(Y = 1) =
0.0051 0.9952 = 0.015
1
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

13 / 1

5. Special random variables

5.2 Binomial random variables

Binomial distribution: examples

Example (Ex. 54 p.54 in the textbook)
Suppose that 10% of all bits transmitted through a digital communication
channel are erroneously received and that whether any is erroneously
received is independent of whether any other bit is erroneously received.
Consider sending a large number of messages, each consisting of 20 bits. a)
What proportion of these messages will have exactly 2 erroneously received
bits? b) What proportion of these messages will have at least 5 erroneously
received bits? c) For what proportion of these messages will more than half
the bits be erroneously received?
Let X be the number of erroneously received bits in a message of 20 bits.
Clearly, we have X Bin(20, 0.1). Thus we have
# $ 2 18
= 0.2852
a) P(X = 2) = 20
2 0.1 0.9

b) P(X 5) = 1P(X = 0)P(X = 1)P(X = 2)P(X = 3)P(X 4) = . . .

c) P(X > 10) = P(X = 11) + P(X = 12) + . . . + P(X = 20) = . . .

very tedious!
CVEN2002/2702 (Statistics)

use statistical tables or statistical software

Dr Justin Wishart

Session 2, 2012 - Week 5

14 / 1

5. Special random variables

5.2 Binomial random variables

Binomial distribution: examples

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

15 / 1

5. Special random variables

5.3 Poisson random variables

The Poisson distribution

Assume you are interested in the number of occurrences of some
random phenomenon in a fixed period of time
Define X = number of occurrences. We say that X is a Poisson
random variable with parameter , i.e.
X P(),
if
p(x) = e

x
x!

for x SX = {0, 1, 2, . . .}

Note: Simeon-Denis Poisson

(1781-1840) was a French
mathematician

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

16 / 1

5. Special random variables

5.3 Poisson random variables

Poisson distribution: how does it arise?

think of the time period of interest as being split up into a large number,
say n, of subperiods
assume that the phenomenon could occur at most one time in each of
those subperiods, with some constant probability
if what happens within one interval is independent to others,
X Bin(n, )
now, as n increases, should decrease (the shorter the period, the less
likely the occurrence of the phenomenon) let = /n for some > 0
then, for any x {0, 1, . . . , n},
! "x !
"nx
n!

P(X = x) =
(binomial pmf)
1
x!(n x)! n
n
!
"n
n!

x
= x
1
n (n x)!(1 /n)x x!
n
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

17 / 1

5. Special random variables

5.3 Poisson random variables

Poisson distribution: how does it arise?

finally, as
n!
1
nx (n x)!(1 /n)x
as n , it remains
P(X = x) = e

and

x
x!

1
n

"n

for x {0, 1, . . .}

which is the Poisson pmf as given on Slide 16

the Poisson distribution is thus suitable for modelling the number of
occurrences of a random phenomenon satisfying some assumptions of
continuity, stationarity, independence and non-simultaneity
is called the intensity of the phenomenon
Note: we defined the P() distribution by partitioning a time period, however
the same reasoning can be applied to any interval, area or volume
the number of flaws in a new windshield might also be Poisson-distributed
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

18 / 1

5. Special random variables

5.3 Poisson random variables

Poisson distribution: pmf and cdf

1.0

lambda = 10

1.0

lambda = 2

1.0

lambda = 1

1.0

lambda = 0.5

1.0

lambda = 0.1

0.8
0.4

pX(x)

0.6

0.8
0.4

pX(x)

0.6

0.8
0.6
0.4

0.4

pX(x)

0.8
0.6

pX(x)

0.8
0.4

pX(x)

0.6

0.2

0.2

0.2

0.2

0.2

15

20

15

20

1.0

1.0

10

10

15

20

0.0

0.0

10

15

20

10

0.8

0.8

0.8
0.6

0.8

0.8

20

0.6

15

1.0

10

1.0

1.0

0.0

0.0

0.0

0.6

FX(x)

0.6

0.4

0.4

FX(x)

0.6
FX(x)
0.4

FX(x)
0.4

0.4

FX(x)

0.2

0.2

0.2

0.2

0.2

10

15

20

10
x

15

20

10

15

20

0.0

0.0

0.0

0.0

0.0

10

15

20

10

15

20

Poisson pmf and cdf, for = {0.1, 0.5, 1, 2, 10}

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

19 / 1

5. Special random variables

5.3 Poisson random variables

Poisson distribution: properties

First we have, as expected,
%

p(x) =

x!

x=0

xSX

=e

%
x
x=0

x!

= e e = 1

Similarly,
E(X ) =

xp(x) =

xe

x=0

xSX

E(X 2 ) =

%
x
x1
e
=
=
x!
(x 1)!
x=1

x 2 p(x) = . . . = 2 +

xSX

Var(X ) =

E(X 2 )

(E(X ))2 = 2 + 2 =

Mean and variance of the Poisson distribution

If X P(),

E(X ) =

CVEN2002/2702 (Statistics)

and
Dr Justin Wishart

Var(X ) =
Session 2, 2012 - Week 5

20 / 1

5. Special random variables

5.3 Poisson random variables

Poisson distribution: examples

Example
Over a 10-minute period, a counter records an average of 1.3 gamma
particles per millisecond coming from a radioactive substance. To a good
approximation, the distribution of the count, X , of gamma particles during the
next millisecond is Poisson distributed. Determine a) , b) the probability of
observing one or more gamma particles during the next millisecond and c)
the variance of this number
a) The mean of the Poisson distribution is , so we can approximate by the
long-run average of the number of particles per millisecond, that is, 1.3.
So we have
X P(1.3)
b) Thus,
1.30
= 1 e1.3 = 0.727
0!
c) The variance of the Poisson distribution is also equal to , hence
P(X 1) = 1 P(X = 0) = 1 e1.3

Var(X ) = 1.3 (particles2 )

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

21 / 1

5. Special random variables

5.3 Poisson random variables

Poisson distribution: examples

Example (Ex. 56 p.55 in the textbook)
Suppose that the number of drivers who travel between a particular origin
and destination during a designated time period has Poisson distribution with
parameter = 20. In the long-run, in what proportion of time periods will the
number of drivers a) be at most 10? b) exceed 20? c) be between 10 and 20,
inclusive? Strictly between 10 and 20?
Let X be the number of drivers. It is given that X P(20)
a)
P(X 10) = P(X = 0) + P(X = 1) + . . . + P(X = 10)
= e20 + e20 20 + e20

202
2010
+ . . . + e20
2
10!

= ...
tedious !
CVEN2002/2702 (Statistics)

use statistical tables or statistical software

Dr Justin Wishart

Session 2, 2012 - Week 5

22 / 1

5. Special random variables

5.3 Poisson random variables

Poisson distribution: examples

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

23 / 1

5. Special random variables

5.3 Poisson random variables

Poisson approximation to the Binomial distribution

Since it was derived as a limit case of the Binomial distribution when n
is large and is small, one can expect the Poisson distribution to be
a good approximation to Bin(n, ) in that case
true
As it involves only one parameter, the Poisson pmf or the Poisson
tables are usually easier to handle than the corresponding Binomial
pmf and tables

Example
It is known that 1% of the books at a certain bindery have defective bindings.
Compare the probabilities that x (x = 0, 1, 2, . . .) of 100 books will have
defective bindings using the (exact) formula for the binomial distribution and
its Poisson approximation
# $
0.01x 0.99100x , while its Poisson
The exact Binomial pmf is p(x) = 100
x
approximation is
x
p (x) = e
!
with = n p = 100 0.01 = 1
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

24 / 1

5. Special random variables

5.3 Poisson random variables

Poisson distribution: examples

Matlab computations give:

We see that the error we would make by using the Poisson

approximation instead of the true distribution is only of order 103
very good approximation
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

25 / 1

5. Special random variables

5.4 Uniform random variables

The Uniform distribution

There are also numerous continuous distributions which are of great
interest. The simplest one is certainly the uniform distribution
A random variable is said to be uniformly distributed over an interval
[, ], i.e.
X U[,]
if its probability density function is given by
, 1
if x [, ]
f (x) =
( SX = [, ])
0
otherwise
Constant density X is just as likely to be close to any value in SX
By integration, it is easy to show that

if x <
0
x
if
x
F (x) =

1
if x >
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

26 / 1

5. Special random variables

5.4 Uniform random variables

0.8

1.0

The Uniform distribution

0.0

0.2

0.4

fX(x)

FX(x)

0.6

pdf f (x) = F (x)

cdf F (x)
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

27 / 1

5. Special random variables

5.4 Uniform random variables

Uniform distribution: properties

Note that the constant Uniform density is set to 1/( ) on [, ] so
-
as to ensure that f (x) dx = 1 (Slide 17, Week 4)
Now,

E(X ) =
Similarly,
2

E(X ) =

/ 2 0
1
1
x
+
2 2
dx =
=
=

2 2( )
2

/ 3 0
1
1
x
2 + + 2
3 3
x
dx =
=
=

3 3( )
3
2

which implies Var(X ) = E(X 2 ) (E(X ))2 = . . . =

()2
12

Mean and variance of the Uniform distribution

If X U[,] ,
E(X ) =
CVEN2002/2702 (Statistics)

+
2

and

Var(X ) =

Dr Justin Wishart

( )2
12
Session 2, 2012 - Week 5

28 / 1

5. Special random variables

5.4 Uniform random variables

Uniform distribution: example

The probability that X lies in any subinterval [a, b] of [, ] is:
P(a < X < b) =

ba

(area of a rectangle)

Example
Buses arrive at a specified stop at 15-minute intervals starting at 7 A . M . That
is, they arrive at 7, 7:15, 7:30, 7:45, etc. If a passenger arrives at the stop at a
time uniformly distributed between 7 and 7:30, find the probability that he
waits less than 5 minutes for a bus
Let X denote the time (in minutes) past 7 A . M . that the passenger arrives at
the stop. We have X U[0,30]

The passenger will have to wait less than 5 min if he arrives between 7:10
and 7:15 or between 7:25 and 7:30. This happens with probability

P((10 < X < 15) (25 < X < 30)) = P(10 < X < 15) + P(25 < X < 30)
5
1
5
+
=
=
30 30
3
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

29 / 1

5. Special random variables

5.5 Exponential random variables

The Exponential distribution

Remind that a Poisson distributed r.v. counts the number of
occurrences of a given phenomenon over a unit period of time
The (random) amount of time before the first occurrence of that
phenomenon is often of interest as well
If N P() denote the number of occurrences over a unit period of
time, then it can be shown that the number of occurrences of the
phenomenon by a time x, say Nx , is P(x)
(Poisson process)
Denote X the amount of time before the first occurrence

This time will exceed x (x 0) if and only if there have been no

occurrences of the phenomenon by time x, that is, Nx = 0
0

x ,
As Nx P(x), it follows P(X > x) = P(Nx = 0) = ex (x)
0! = e
which yields the cdf of X :

F (x) = P(X x) = 1 ex

for x 0

This particular distribution is called the Exponential distribution

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

30 / 1

5. Special random variables

5.5 Exponential random variables

The Exponential distribution

A random variable is said to be an Exponential random variable with
parameter ( > 0), i.e.
X Exp(),

if its probability density function is given by

, x
e
if x 0
f (x) =
( SX = R+ )
0
otherwise
By integration, we find
F (x) =

0
1 ex

if x < 0
if x 0

From the preceding argument, it is easy to understand why this

distribution is often useful for representing random amounts of time,
like the amount of time required to complete a specified task, the
waiting time at a counter, the amount of time until you receive a phone
call, the amount of time until an earthquake occurs, etc.
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

31 / 1

5. Special random variables

5.5 Exponential random variables

The Exponential distribution

1/2

fX(x)

FX(x)

ln 2
0

pdf f (x) = F (x)

cdf F (x)
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

32 / 1

5. Special random variables

5.5 Exponential random variables

Exponential distribution: properties

We can check that (as expected)
/ x 0+
. +
. +
e
x
f (x) dx =
e
dx =
=1
() 0

0
Moreover,
E(X ) =

x e

dx = x

/ x 0+
e
1
=0+
=
0

Similarly, E(X 2 ) =

- +
0

2+
ex 0

x 2 ex dx = . . . =

ex dx

(by parts)

2
,
2

so that Var(X ) =

1
2

Mean and variance of the Exponential distribution

If X Exp(),
E(X ) =
CVEN2002/2702 (Statistics)

and
Dr Justin Wishart

Var(X ) =

1
2
Session 2, 2012 - Week 5

33 / 1

5. Special random variables

5.5 Exponential random variables

Exponential distribution: example

Example
Suppose that, on average, 3 trucks arrive per hour to be unloaded at a
warehouse. What is the probability that the time between the arrivals of two
successive trucks will be a) less than 5 minutes? b) at least 45 minutes?
Assuming the number of trucks arriving during one hour is Poisson distributed
(with parameter = 3), then the amount of time X between two truck arrivals
follows the Exp(3) distribution
Hence,
- 1/12 3x
a) P(X 1/12) = 0
3e
dx = 1 e1/4 = 0.221
-
b) P(X > 3/4) = 3/4 = e9/4 = 0.105

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

34 / 1

5. Special random variables

5.5 Exponential random variables

Other useful distributions

In the remainder of this course we will also encounter some other

continuous distributions, among these are
the Student-t (or just t) distribution, X t ;
the 2 distribution, X 2 ;

the Fisher-F (or just F ) distribution, X Fd1 ,d2

We will return to them later when we will need them

The several distributions that we have introduced so far are very useful
in the application of statistics to problems of engineering and physical
science

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

35 / 1

5. Special random variables

5.6 Normal random variables

The Normal distribution: introduction

However, the most widely used, and therefore the most important,
statistical distribution is undoubtedly the
Normal distribution
Its prevalence was first highlighted when it was observed that in many
natural processes, random variation among individuals systematically
conforms to a particular pattern:
most of the observations concentrate around one single value
(which is the mean)
the number of observations smoothly decreases, symmetrically on
either side, with the deviation from the mean
it is very unlikely, yet not impossible, to find very extreme values
this yields the famous bell-shaped curve
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

36 / 1

5. Special random variables

5.6 Normal random variables

The Normal distribution: introduction

The bell-shaped curve was first spotted by the French mathematician
Abraham de Moivre (1667-1754) who in his 1738
#n$book The Doctrine
n
of Chances showed that the coefficients Ck = k in the binomial
expansion of (a + b)n (see Slide 7) precisely follow the bell shape
pattern when n is large
n= 10
250

10

n= 5

100
50

Cnk

Cnk

150

200

n= 20

2.0e+299
Cnk

50000

1.0e+299

Cnk

100000

10

15

20

0.0e+00

150000

10

400

CVEN2002/2702 (Statistics)

n=1000

6
k

450

500

550

600

Dr Justin Wishart

Session 2, 2012 - Week 5

37 / 1

5. Special random variables

5.6 Normal random variables

The Normal distribution: introduction

Later, Carl-Friedrich Gauss (1777-1855), a German mathematician
(sometimes referred to as the Princeps mathematicorum, Latin for "the Prince
of Mathematicians" or "the foremost of mathematicians") , was the first to
write an explicit equation for the bell-shaped curve:
1

2
1
(x) = ex /2
2

(x)

When deriving his distribution, Gauss was primarily interested in errors of

measurement, whose distribution typically follows the bell-shaped curve as
well. He called his curve the normal curve of errors, which was to become
the Normal distribution. In honour of Gauss, the Normal distribution is also
often referred to as the Gaussian distribution
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

38 / 1

5. Special random variables

5.6 Normal random variables

The Normal distribution: introduction

It is important to note that the Normal distribution is not just a
convenient mathematical tool, but also occurs in natural phenomena
For instance, in 1866 Maxwell, a Scottish physicist, determined the
distribution of molecular velocity in a gas at equilibrium. As a result of
unpredictable collisions with other molecules, molecular velocity in a
given direction is randomly distributed, and from basic assumptions,
that distribution can be shown to be the Normal distribution

James Maxwell (1831-1879)

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

39 / 1

5. Special random variables

5.6 Normal random variables

The Normal distribution: introduction

Another famous example is the bean machine, invented by Sir
Francis Galton (English scientist, 1822-1911) to demonstrate the
Normal distribution. The machine consists of a vertical board with
interleaved rows of pins. Balls are dropped from the top, and bounce
left and right as they hit the pins. Eventually, they are collected into
bins at the bottom. The height of ball columns in the bins
approximately follows the bell-shaped curve

CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

40 / 1

5. Special random variables

5.6 Normal random variables

The Normal distribution

A random variable is said to be normally distributed with parameters
and ( > 0), i.e.
X N (, ),
if its probability density function is given by
f (x) =

1
2

(x)2
2 2

( SX = R)

Unfortunately, no closed form exists for

. x
(y )2
1

F (x) =
e 22 dy
2
Important remark: Be careful! Some (many?) sources use the
alternative notation
X N (, 2 )

in the textbook and in Matlab, the notation N (, ) is used, so we

adopt it in these slides as well
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

41 / 1

5. Special random variables

5.6 Normal random variables

The Normal distribution

fX(x)

FX(x)

22

1/2

0
2

+ 2

+ 2

pdf f (x) = F (x)

cdf F (x)
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

42 / 1

Objectives

Objectives
Now you should be able to:
Understand the assumptions for some common discrete
probability distributions
Select an appropriate discrete probability distribution to calculate
probabilities in specific applications
Calculate probabilities, determine means and variances for some
common discrete probability distributions
Understand the assumptions for some common continuous
probability distributions
Select an appropriate continuous probability distribution to
calculate probabilities in specific applications
Calculate probabilities, determine means and variances for some
common continuous probability distributions
Recommended exercises Q53 p.54, Q55 p.55, Q57 p.55, Q33
p.221, Q37 p.222, Q23 p.78, Q19 p.31, Q23 p.31, Q62 p.56
CVEN2002/2702 (Statistics)

Dr Justin Wishart

Session 2, 2012 - Week 5

43 / 1