Topic 5

DISCRETE DISTRIBUTION
 1. The Discrete Uniform
Distribution
• Equal (uniform) probability for all x values.

• The probability distribution:

1
f(x) = , x  x1 , x2 ,...xk
k k
1
• Mean,  = E(X) =  xi .
i 1 k
k
1
• Variance, var(X) =  ( xi   ) .
2

i 1 k

2
 1. The Discrete Uniform
Distribution
• The probability distribution graph ,f(x),
Pr

xi

Example 1:
Let X be the number of points when we roll a balanced die,
1
f(x) = , x 1, 2,3, 4,5,6
6 3
 2. The Bernoulli Distribution,
X ~ Ber() or X ~ Bin(1, )
 There’s only one trial/experiment.
 The experiment has 2 outcomes,

‘success’ ‘failure’
 Pr(success) = , Pr(failure) = (1 - )
 X: number of success,
x = 0 or 1.
 The probability distribution,
f ( x)   (1   ) ,
x 1 x
x  0,1
4
 2. The Bernoulli Distribution,
X ~ Ber() or X ~ Bin(1, )
 Example of Bernoulli cases,
 Win or Lose,

 Dead or Alive,

 Defective or Non-Defective.

Example 2: Tossing a fair coin

X: Number of Tail obtained

 
x
f ( x)  1 ( 1 )1 x
2 2
x = 0, getting Head,
x = 1, getting Tail.
5
 3. The Binomial Distribution,
Y ~ Bin (n, )
Conditions of a Binomial Experiment
A binomial experiment must satisfy the
following four conditions.
1. There are n identical trials.
2. Each trail has only two possible outcomes.
3. The probabilities of the two outcomes remain
constant.
4. The trials are independent.

6
 The Binomial Probability
Distribution and Binomial Formula
 For a binomial experiment, the probability of exactly x successes
in n trials is given by the binomial formula


n x
P ( x)  n C x p q x

 where
 n = total number of trials
 p = probability of success
 q = 1 – p = probability of failure
 x = number of successes in n trials
 n - x = number of failures in n trials
n y
f ( y )    (1   ) n y , y  0,1,..., n
 y 7
 Mean and Standard Deviation
of the Binomial Distribution
 The mean and standard deviation of a
binomial distribution are

  np and   npq
 where n is the total number of trails, p is the
probability of success, and q is the
probability of failure.

8
 Mean and Standard Deviation
of the Binomial Distribution
Find the mean and variance using
(a) Expectation definition:
(Freund, page 168)
E(X) =
 x. f ( x)

E(X2) = E[X(X – 1)] + E(X)

= + (n)
 x( x  1). f ( x)
var(X) = [n(n – 1)2 + (n)] - (n)2
= n(1 - )

9
 Mean and Standard Deviation
of the Binomial Distribution
Find the mean and variance using
(a) Moment Generating Function (MGF)
 Mgf for Binomial Distribution,

M X (t )  [ et  (1   )]n
 Prove: (Freund, page 171)
E(X) = M x ' (t ) t 0 E ( X 2 )  M X " (t ) t 0


 n e  (1   )
t
 n 1
.e t
t 0
 ......  n(n  1) 2  n

 n  (1   ) .  n
n 1

Var ( X )  E ( X 2 )  [ E ( X )]2  n (1   ) 10
 4. The Poisson
Distribution, X ~ Po()
 When the interest is to find the number of
random occurrences in an interval/certain
space.
 Examples of interval/space:
 Time, volume, area, length, etc.
 Examples of cases:
 Number of telephone calls in an hour
 Number of typo errors in a page
 Number of insects in a plot of land
 Number of road accidents in a year
 Number of demands for a certain good in a week.
11
 THE POISSON DISTRIBUTION
cont.
 Conditions to Apply the Poisson Probability
Distribution

 The following three conditions must be satisfied to

apply the Poisson probability distribution.
1. x is a discrete random variable.
2. The occurrences are random.
3. The occurrences are independent.

12
 THE POISSON DISTRIBUTION
cont.
 Poisson Probability Distribution Formula

 The probability of x occurrences in an interval is

x e  
P( x)  , x=0, 1, 2, …
x!
 where λ is the mean number of occurrences in that
interval and the value of e is approximately
2.71828.
13
Theorem
 
  2

 
t
 ( e 1)
M X (t )  e
14
 Binomial distribution approximation
to Poisson distribution

Theorem: A r.v. X, Binomially distributed will

approximate a Poisson distribution
when n→∞, θ→ 0 i.e.,
- n ≥ 20
- θ ≤ 0.05

 X ~ Bin (n, θ) → Y ~ Po(  = nθ)

E(Y) = var(Y) =  = nθ

15
5. The Geometric
Distribution, X ~ Geo( )

16
Example 12

17
 5. The Geometric
Distribution, X ~ Geo( )
Definition

18
6. The Negative Binomial
Distribution, X ~ BN(r,)
 X: No. of attempts needed to get the rth success,
 x  1 r x r
f ( x)     (1   ) x=r, r+1, r+2,…
 r  1 
 Mean, E(X) = r

r (1   )
var(X) = 2

t r
 e 
M X (t )   t 
 1  (1   ) e 

19
 6. The Negative Binomial
Distribution, X ~ BN(r,)

Negative binomial random variable represented as a

sum of geometric random variables.
20
 6. The Negative Binomial
Distribution, X ~ BN(r, )
 The differences between Binomial distribution and
Negative Binomial distribution

Binomial Negative Binomial

 n trial fixed  r success fixed
 X: no. of successes  X: no. of trials
How many successes How many trials to get
in n trials. r successes

 In the literature of statistics, negative Binomial

distributions are also referred to as binomial
waiting-time distribution or a Pascal
distribution. 21
 6. The Negative Binomial
Distribution, X ~ BN(r, )
Example:
 A thief got caught for the second time in the 8th

attempt to steal
 The 8th child born is the 4th baby boy of the family

 The 6th child exposed to a certain contagious

disease is the third to catch it.

_S_S_S
(r-1) success at the rth success
(x-1) trial at the xth trial

22
6. The Negative Binomial
Distribution, X ~ BN(r, )

23
7. The Hypergeometric
Distribution, X ~Hpg(n, N,m)

24
 7. The Hypergeometric
Distribution, X ~Hpg(n, N,m)
 For the case of sampling without replacement and when the
trials are not independent.
 For example, there are N lamps and m of them are defective.
Taking n samples of the lamp, it is found that x of the lamps are
defective.

m: (success)
N
N-m: (failure) We are interested
in x successes in
n trials
x: (success)
n
n-x: (failure)
25
 7. THE HYPERGEOMETRIC
DISTRIBUTION
 The probability of x successes in n trials is given by

C x N  r Cn  x
P( x)  r

N Cn
or,
 m  N  m  x = 0, 1, ..., n
 x  n  x 
f ( x)    , xm
N
n  (n-x)  (N-m)
 
X: number of success
N: number of population
n: sample size
m/r: number of elements of interest 26
 7. Hypergeometric Distribution

Mean and Variance

27
 Hypergeometric approximation
to Binomial Distribution
 Rule for approximation:
n
 5%
N
 m
X~Hpg(n, N, m)  Y~Bin n,  
 N

28