CSC 124 Lec 9 Probability and Statistics

1 Discrete Probability Distribution Functions

1.1 Poisson Distribution Function
Distribution due to Simeon D. Poisson (1781 - 1840)
Definition:
A random variable X is defined to have a Poisson distribution if its p.d.f is
given by

e−λ λx

x! , x = 0, 1, · · ·
f (x) =
0, elsewhere
where the parameter λ satisfies λ > 0.
Note:
We know that
∞
X λx
eλ = ,
x=0
x!
hence
∞ ∞ ∞
X X e−λ λx X λx
f (x) = = e−λ = e−λ eλ = 1
x=0 x=0
x! x=0
x!
Theorem
Let X be a Poisson distributed random variable. Then E(X) = λ, V ar(X) =
t
λ and M (t) = eλ(e −1)
Proof:
The moment generating function, mgf , denoted as M (t), by definition

1
X
M (t) = E(etx ) = etx f (x) (1)
x=0
n
X e−λ λx
= etx
x=0
x!
∞ x
X (λet )
= e−λ
x=0
x!
t t
= e−λ eλe = eλ(e −1)

1
 Then, the mean and variance from the moment generating function, M (t),
becomes
Mean,

d −λ λet
M (1) (t) = e e (2)
dt t=0
d  t
= e−λ eλe
dt t=0
t d t
e−λ eλe λet = λe−λ et eλe


=
dt t=0 t=0
(1)
=⇒ E(X) = M (t = 0)
= λe−λ eλ
= λ

For the variance, we get the 2nd moment from the mgf as

d2 t
M (2) (t) = 2
λe−λ et eλe (3)
dt  t=0

−λ t λet t
= λe e e + et λet eλe
t=0
t
λe−λ et eλe 1 + λet


=
t=0
⇒ E(X 2 ) = M (2) (t = 0)
= λ + λ2

Thus,

 2
V ar(X) = M (2) (t = 0) − M (1) (t = 0) (4)
2
= E X 2 − (E(X))

= λ + λ2 − λ2
= λ

Alternatively, without making use of the mgf ;

By definition,

n
X
E(X) = xf (x) (5)
x=0
∞
X e−λ λx
= x
x=0
x!

2
 ∞
X e−λ λx
= x
x=0
x!
∞
X λx−1
= e−λ λ
x=1
(x − 1)!
| {z }
Sums to eλ
= λ

and by letting X 2 = X(X − 1) + X, then

n
X
E(X 2 ) = x2 f (x) (6)
x=0
∞
X e−λ λx
= [x(x − 1) + x]
x=0
x!
∞ ∞
X e−λ λx X e−λ λx
= x(x − 1) + x
x=2
x! x=1
x!
∞ ∞
X λx−2 X λx−1
= e−λ λ2 +e−λ λ
x=2
(x − 2)! x=1
(x − 1)!
| {z } | {z }
Sums to eλ Sums to eλ
2
= λ +λ

leading to

2
E X 2 − (E(X))


V ar(X) = (7)
2 2
= λ +λ−λ
= λ

as obtained in 4 above.

1.1.1 Examples
1. The number of deaths by suicide in a certain institution is distributrd as
a poisson process with parameter λ = 2 per day. Let X be the number
of deaths by suicide per day. Find (i) P (X = 2) (ii) P (X = 0) (iii)
P (X ≥ 3). Solution:

e−2 22
P (X = 2) = = 2e−2 = 0.2707
2!
e−2 20
P (X = 0) = = e−2 = 0.1353
0!

3
 P (X ≥ 3) = P (X = 3) + P (X = 4) + · · ·
= 1 − {P (X = 0) + P (X = 1) + P (X = 2)}
= 1 − (0.1353 + 0.2707 + 0.2707)
= 0.3233

2. If the average number of deaths per day is λ = 2, then the average number
of deaths per week λ1 = 7λ = 7 × 2 = 14 deaths per week. If we let X to
be the number of deaths per week, then
 −14 x
e 14
, x = 0, 1, 2, · · ·
f (x) = x!
0, elsewhere

Find P (X = 0), P (X = 1), P (X = 2) and P (X = 3). Solution:

e−14 140
P (X = 0) = = e−14 = 0.000000831
0!
e−14 141
P (X = 1) = = 14e−14 = 0.00001164
1!
e−14 142
P (X = 2) = = 7 × 14e−14 = 0.0000815
2!
e−14 143 143 −14
P (X = 3) = = e = 0.00038
3! 6
3. A controlled manufacturing process is 0.1% defective. What is the prob-
ability of taking 3 or more defectives from a lot of 100 pieces when you
use, (i) Binomial distribution (ii) Poisson distribution. Solution: (i) p =
Probability of defective = 0.002, q = 1 - p = 0.998 and n = 100
P (3 or more def ectives) = P (3) + P (4) + · · · + P (100)
⇒ P (x
≥ 3) = 1− {P (x = 0) + P (x
 = 1) +
 P (x = 2)}   
100 0 100 100 1 99 100 2 98
=1− 0.001 0.999 + 0.001 0.999 + 0.001 0.999
0 1 2
= 1 − {0.9048 + 0.09048 + 0.004488}
= 1 − 0.999768
= 0.000232

(ii) Poisson: Probability of finding 2 or more defective

Here, λ = np
0.1
= 100 × 100 = 0.1

P (3 or more def ectives) = P (3) + P (4) + · · · + P (100)

P (x ≥n3) = 1 − {P (x = 0) + P (x = 1)o+ P (x = 2)}
−0.1 −0.1
0 1
e−0.1 0.12
= 1 − e 0!0.1 + e 1!0.1 + 2!
= 1 − e−0.1 {1 + 0.1 + 0.005}
= 1 − 0.99985
= 0.00015

4
2 Continuous Probability Distribution Functions
• Uniform distribution functions

• Exponential distribution function

• Normal distribution functions

2.1 Uniform distribution function

Definition:
A random variable X is defined to have a Uniform distribution if the prob-
ability distribution function of X is given by
 1
f (x) = b−a , a < b
0, elsewhere

In this case, X is defined over the interval (a, b).

Theorem:
(b−a)2
If X has a Uniform distribution, then E(X) = a+b 2 , V ax(X) = 12 and
etb −eta
the mgf is Mx (t) = t(b−a)
Proof:
R∞ 1
Rb x
E(X) = x b−a dx = b−a dx
 −∞  a
b
1 x2 a+b
= b−a 2 = 2
x=a

Rb x2
E(X 2 ) = b−a dx
 a 
b
1 x3 b2 +ab+a2
= b−a 3 = 3
x=a

⇒ V ar(X) = E(X 2 ) − (E(X))2

2 2
2 2
= b +ab+a
3 − a+b
2 = (b−a)
12

and
Rb 1
Mx (t) = E(etx ) = etx b−a dx
  a
b
1 etx tb
−eta
= b−a t = et(b−a)
x=a

5
2.2 Exponential distribution function
Definition:
A random variable X is defined to have an Exponential distribution if the
probability distribution function of X is given by
( x
1 − /β
f (x) = βe ,β > 0
0, e; sewhere

Theorem:
If X has an Exponential distribution, then E(X) = β, V ar(X) = β 2 and
1
the mgf is Mx (t) = 1−βt , t < β1
Proof:
R∞ −x/β
E(etx ) = etx β1 e dx
0
R∞ tx−x/
= 1
e β dx
β
0
" #∞1
1 e β
(t− / )x
= β 1
t− /β
x=0
= −1
βt−1
1
= 1−βt ; t < 1/β

Note:
1
If we let β = λ from the above given Exponential distribution, then the
function
λe−λx , x > 0

f (x) =
0, elsewhere
also represents an Exponential distribution where λ is a positive constant.
Where the mgf is
R∞
E(etx ) = etx λe−λx dx
0
R∞ −x(λ−t)
=λ e dx
0
h −x(λ−t)
i ∞
e
=λ λ−t
x=0
λ
= λ−t ; t<λ
with both the expectation and the variance being λ1 and λ12 respectively.
Exercise:
In both cases, show the expectation and variance, both from definition and
from the mgf.