DISTRIBUTIONS

Special discrete probability distributions

Bernoulli distribution
A random variable X is defined to have a Bernoulli distribution if the discrete
density function of X is given by

{
x 1−x
f ( x )= p ( 1− p ) ,∧x=0 , 1 ,q=1−p
0 ,∧otherwise

The mean of X is given by

1
E ( X )=∑ x f (x )
x=0

1
¿ ∑ x px ( 1− p )
1− x
=p .
x=0

The variance of X is given by

2
Var ( X )=E ( X )− [ E ( X ) ]
2

But E ( X 2) = ∑ x 2 f ( x )
x=0

1
¿ ∑ x 2 p x ( 1− p )
1−x
=p
x=0

2
Therefore Var ( X )=E ( X 2 )− [ E ( X ) ] = p− p2 =pq .
Example 7.1
A box contains 4 good fruits and 6 bad ones. If a fruit is selected at random from
the box, it can either be good or bad. The random variable
 {
X = 1 ,if the fruit is bad
0 ,if the fruit is good

is a Bernoulli random variable. The probability of selecting a good fruit is the

proportion of good fruits in the box. Hence
4 2
p=P ( X =1 )= = .Thus
10 5

E ( X )=p=
2
5
and r ( X ) =pq=
2
5
2
( )( )
6
1− = .
5 25

The moment generating function of a Bernoulli random variable is given by

1
M X ( t )=E ( e tX ) =¿ ∑ etx f (x )
x=0

1
¿ ∑ etx p x ( 1− p )
1−x

x=0

t
¿ ( 1− p )+ p e

Which exists for all real numbers t.

We can obtain the mean and variance of X by using the above m.g.f as follows:
M X ( t )= p e .Therefore
∕ t

∕
E ( X )=M X ( 0 )= p .

M X ( t )= p e . Thus
∕ ∕ t

M X∕ ∕ ( 0 )= p=E ( X 2 ). Hence
2
Var ( X )=E ( X )− [ E ( X ) ] = p− p =pq .
2 2

BINOMIAL DISTRIBUTION
A random variable X is defined to have a binomial distribution if the discrete
density is given by
 {( )
x
n p ( 1− p )n− x ,∧x=0 , 1, … , n , q=1− p
f ( x )= x
0 ,∧otherwise

The parameters of the above distribution are n and p, where 0 ≤ p ≤ 1.

The mean of X is given by
n
E ( X )=∑ x f ( x )
x=0

n x

x=0 x ()
¿ ∑ x n p (1−p ) =np.
n−x

n
E ( X 2) = ∑ x 2 f ( x )
x=0

n x

X =0 x ()
¿ ∑ x2 n p (1−p ) =n ( n−1 ) p2 +np.
n−x

Therefore the variance of X is given by

2
Var ( X )=E ( X )− [ E ( X ) ] =n ( n−1 ) p +np−n p =npq .
2 2 2 2

Example 7.2
In a certain community, the probability of a female birth is 0.3 . If ten individuals
are randomly selected from this community, calculate
(i) the probability that exactly six of them are males,
(ii) the average number of females in the sample. Assume that the rate of
survival is the same for both sexes.
Solution
Let X denote the number of females in the sample. Then assuming the binomial
distribution, we have

{( )
x
10 ( 0.3 ) ( 0.7 )10− x ,∧x=0 , 1, … , 10
P ( X )=f ( x )= x
0 ,∧otherwise

(i) Prob ( 6 males )=P (X=10−6)

 ¿ P(X =4 )
4

( )
¿ 10 ( 0.3 ) ( 0.7 ) =0.2001
4
10−4

(ii) E ( X )=np=10× 0.3=3

The moment generating function of the binomial random variable X with

parameters n and p is given by
n
M X ( t )=E ( e ) =¿ tX
∑ etx f (x )
x=0

x=0 x ()
¿ ∑ etx n p x ( 1− p )
n− x

()
¿ ∑ n ( p e ) ( 1−p ) =( p e + q )
t x n−x t n

x=0 x

The mean and variance of a binomial random variable X are determined by using
its m.g.f as follows:
n−1
M X ( t )=np e ( p e +q )
∕ t t
. Therefore
∕
E ( X )=M X ( 0 )=np ( p+ q )
n−1
.
¿ np , since p+q=1.
t 2 n−2 n−1
M X ( t )=n ( n−1 ) ( p e ) ( p e +q )
¿∕ t
+ np e ( p e + q )
t t
.
E ( X 2) =M ¿X∕ ( 0 )=n ( n−1 ) p 2 ( p+ q )
n−2 n−1
+np ( p+q )
2
¿ n ( n−1 ) p +np
2
Var ( X )=E ( X )− [ E ( X ) ] =n ( n−1 ) p +np−n p =npq .
2 2 2 2

Geometric Distribution
A random variable X is defined to have a geometric distribution if the p.d.f. of X is given by

{
x−1
p ( 1− p ) , x = 1 , 2, …
f (x) =
0 , elsewhere
Mean and Variance.

We use the m.g.f. technique

M x (t ) = E ( e tx )
∞
= ∑ e tx f ( x )
x=0
∞
= ∑ etx p ( 1− p ) x−1
x=1
∞ x −1
=p ∑ [ e t ( 1− p ) ]
x=1

[ 2
=p 1+ e t ( 1− p )+ ( e t ( 1− p ) ) + … ]
p
=
1−e t (1− p )
p
= , where q=1−p
1−qe t

pqe t
M 'x (t ) =
( 1−qet )2
E ( x ) = M 'x ( 0 )
pq q
= 2
=
(1−q ) p

2 pq 2 e2 t pqet
M ''x (t ) = +
( 1−qet )3 ( 1−qet )2
M ''x ( 0 )=
2 pq 2 pq
= +
(1−q ) ( 1−q )2
3

2q 2 q
= 2 +
p p
Var ( x ) = M ''x ( 0 ) −[ M 'x ( 0 ) ]
2

2 q2 q
()
2
q
= 2
+ −
p p p
q2 q
= +
p2 p
q2 + pq
= 2
p
q ( p+q )
= 2
p
q
= 2
p

Assessment
1. A study has shown that 80% of all families living in a certain residential
estate in Nakuru own a TV set. If 20 families are randomly selected from
this estate, compute the probability that
(i) all will have TV sets,
(ii) between 8 and 10, inclusive will have TV sets,
(iii) at most 10 will have TV sets,
(iv) at least 15 will own TV sets.
2. Suppose X is binomially distributed with parameters n and p; further
suppose that E(X)=5 and Var(X)=4. Find the values of n and p.
 POISSON DISTRIBUTION
Poisson distribution
A random variable X is defined to have a Poisson distribution if its density is given
by

{
e−λ λ x
P ( X=x ) =f ( x )= x ! ,∧x=0 , 1 ,2 , …
0 ,∧otherwise 0

Where the parameter λ satisfies λ> 0.

The mean of Poisson random variable X is given by
∞
E ( X )=∑ xf ( x )
x=0

∞
e−λ λ x
¿∑ x
x=0 x!
∞
λ x−1
¿ e− λ λ ∑ x
x=1 x (x−1)!

¿ e λ e =λ .
−λ λ

The variance of X is given by

2
Var ( X )=E ( X )− [ E ( X ) ] .
2

But E ( X 2) = ∑ x 2 f ( x )
x=0

∞
e−λ λ x
¿ ∑ x2
x=0 x!
∞
e− λ λ x
¿ ∑ [x ( x−1 ) + x ]
x=1 x!
∞ ∞
e− λ λ x e−λ λx
¿ ∑ x ( x−1 ) +∑ x
x=1 x ! x=0 x!
∞ ∞
e− λ λ x e−λ λ x
¿ e λ ∑ x ( x −1 )
−λ 2
+∑ x
x=2 x (x−1)( x−2) ! x=0 x!
−λ 2 λ
¿e λ e + λ
 ¿ λ + λ.
2

∴ Var ( X )= λ2 + λ−λ2= λ.

Example 8.1
The number of male mates of a queen bee was found to have a Poisson
distribution with parameter λ=2.7. Find the probability that the number, X, of
male mates of a queen bee is
(i) exactly 2,
(ii) at most 2,
(iii) between 1 and 3, inclusive,
Solution
The probability distribution of X is given by
e−λ λ x
{
P ( X=x ) =f ( x )= x ! ,∧x=0 , 1 ,2 , …
0 ,∧otherwise 0
2
−2.7 2.7
(i) P ( X=2 )=f ( 2 ) =e =0.2450
2!
(ii) P ( X ≤2 )=P ( X=0 )+ P ( X=1 ) + P( X=2)
0 1 2
−2.7 2.7 −2.7 2.7 −2.7 2.7
¿e +e +e
0! 1! 2!
= 0.0672+0.1815+0.2450 = 0.4937
1 2 3
2.7 −2.7 2.7 −2.7 2.7
(iii) P ( 1≤ X ≤ 3 )=e
−2.7
+e +e =0.6470
1! 2! 3!

The moment generating function of the Poisson random variable X with

parameter λ given by
M X ( t )=E ( e tX )
∞
¿ ∑ e f (x)
tx

x=0

∞
e− λ λ x
¿ ∑ etx
x=0 x!
∞ t x
(λe )
¿ e− λ λ ∑
x=0 x!
 t

¿ e− λ e λ e

¿ e λ (e −1) .
t

The mean and variance of X can be obtained using this m.g.f as follows:
M X∕ ( t )=λ e t e λ (e −1) .
t

∴ E( X ) = M X∕ ( 0 ) =λ

M X∕ ∕ ( t )= λ2 e2 t e λ (e −1) +¿ λ e t e λ (e −1)
t t

Therefore E ( X 2) =M X∕ ∕ ( 0 )=λ 2 e 0 e λ ( e −1 )+ ¿ λ e 0 e λ (e −1)=λ 2+ λ.

0 0

2
Thus Var ( X )=E ( X )− [ E ( X ) ] =λ + λ−λ =λ .
2 2 2

Approximating Binomial Probabilities

The Poisson distribution can be used in approximating binomial probabilities
when the number of trials n becomes large, and the expected number of
successes np remains unchanged.
Consider the binomial distribution with parameters n and p. Then

{( )
x
n p ( 1− p )n− x ,∧x=0 , 1, … , n , q=1− p
( )
f x= x
0 ,∧otherwise

m
Let m=np (constant) ⇒ p= n . Then

( )( )
x n− x
n! m m
f ( x )= 1−
x ! (n−x)! n n

n ( n−1 ) ( n−2 ) …(n−x+1)(n−x)! m x

( )( )
n −x
m m
¿ 1− 1−
x !(n−x )! n
x
n n

( ) ( )( )
n −x
1 n n−1 n−x +1 x m m
¿ × ×…× m 1− 1−
x! n n n n n

But
 ( ) ( ) ( )
−x ❑ n
m n n−1 n−x+1 m −m
lim 1− =1 , lim × ×…× =1 and lim 1− =e .
n→∞ n n→∞ n n n n→∞ n
Therefore taking limits as n → ∞ and holding np fixed,
We have
x −m
m e
f (x)→ . Hence
x!

−np x
e ( np )
n→∞ x ()
lim n p (1−p ) =
x n−x
x!

for fixed np. Thus for large but finite n and small p, one can approximate the
binomial distribution with parameters n and p with the Poisson distribution with
mean m=np .
Example 8.2
A machine produces 1% defective items. Suppose it produces 1000 items. What is
the probability that an item selected at random is defective?
Solution
Let X be the number of defective items among the 1000 items produced by the
machine. Then X is binomially distributed with parameters
n=1000 and p=0.01. Therefore

{(
x

f ( x )= x )
1000 (0.01) ( 0.99 )1000− x ,∧x=0 ,1 , … , 1000

0 ,∧otherwise
1

1 ( )
P ( X=1 )=f ( 1 )= 1000 (0.01) ( 0.99 ) =0.00044
999

Using the Poisson approximation with m=np=10 we have

x −10 x

( )
1000 (0.01) ( 0.99 )1000−x ≃ e 10
x x!
And
 −10 1
e 10
P ( X=1 ) ≃ =0.0005
1!

Assessment
1. Use the Poisson approximation to compute the following probabilities
(i) P ( X=45 ) , where X is a binomial random variable with parameters n=100
and ¿ 0.5 .
(ii) P ( X ≤2 ) , where X is a binomial random variable with parameters n=120
and ¿ 0.04 .
2. If X is a random variable with Poisson distribution satisfying P ( X=0 )=P ( X =1 ) ,
what is (X ) ?
1
3. If X has a Poisson distribution and P ( X=0 )= 2 , what is (X ) ?

Normal Distribution
A continuous random variable X is defined to be normally distributed if its density
is given by
 {
−1 2
( x−μ )
1 2σ
2

f ( x )= e ,∧−∞ < x <∞ ,−∞< μ< ∞ , σ >0

√ 2 π σ2
0 ,∧otherwise

The graph of the normal distribution, called the normal curve, is a belled-shaped
curve that extends indefinitely in both directions, with the horizontal axis as its
asymptote.
If a random variable X is normally distributed with mean μ and standard deviation
σ , then it is usual to write

X ∽ N (μ,σ)

Moments of Normal Distribution

Suppose that a random variable X is normally distributed with mean μ and
standard deviation σ . Then the moment generating function of X can be obtained
as follows:
M X ( t )=E ( e tX )

Standardizing X we have
X−μ
Z=
σ

So that Z ∽ N ( 0 , 1 ). But the moment generating function (m.g.f) is given by

2
t
M Z ( t )=E ( e ) =e ,−∞ <t <∞ .
tZ 2

Now X =σZ + μ .Thus

tX
M X ( t )=E[e ]
t ( σZ+μ )
¿ E [e ]

¿ e tμ E [ e tσZ ]

Putting t ⋇=tσ we have

∴ M X ( t )=e tμ E [ e t Z ]
¿

¿2
t
tμ 2
¿e e
 1 2 2
tμ+ σ t
2
¿e

and it exists for all real numbers t.

We can compute the mean and variance of X by using the above m.g.f.
1 2 2
tμ+ σ t
M X∕ ( t )=( μ+ σ 2 t ) e 2

∴ E ( X )=M X ( 0 )=μ e =μ.

∕ 0

1 2 2 1 2 2
2 tμ+ σ t tμ+ σ t
M X ( t )=( μ+ σ t ) e
∕ ∕ 2 2 2 2
+σ e
2
∴ E ( X ) =M X ( 0 ) =μ + σ ,∴ Var ( X ) =E ( X ) −[ E ( X ) ]
2 ∕ ∕ 2 2 2

Var ( X )=μ + σ −μ =σ .
2 2 2 2

Example 9.1
A random variable X is normally distributed with mean μand variance σ 2 .
Determine the mean and variance of a new random variable Y =e X .
Solution
E ( Y )=E ( e X )=M X ( 1 )
1 2 2
Where M X ( t )=e
tμ+ σ t
2 is m.g.f of X.
1 2
μ+ σ
2
∴ E ( Y )=e
2
Var ( Y )=E ( Y ) −[ E ( Y ) ] .
2

E ( Y 2 ) =E ( e 2 X )=M X ( 2 ) =e 2 μ+2 σ
2

But
2 2

Therefore Var ( Y )=e 2 μ +2 σ −e2 μ+σ .

Computing Normal Probabilities

If X ∽ N ( μ , σ ). Then P ( a ≤ X ≤b )=F ( b )−F ( a ) . Therefore to compute this probability
we need to standardize X i.e
X−μ
Z=
σ
Where Z ∽ N ( 0 , 1 ). The p.d.f of Z is given by

{
−1
1
2
z

f ( z )= √ 2 π e 2
,∧−∞ < z <∞
0 ,∧otherwise

The standardized normal curve is symmetrical about z=0. Thus

P ( Z ≤−a )=P ( Z ≥ a )=1−P(Z ≤ a)

For any real number a (see the figure below)

Graph of f(z)
z −1 2
1 y
Φ ( z )= ∫ e 2
dy . Therefore
−∞ √2 π
Φ (−a )=1−Φ ( a ) or more generally

Φ (−z )=1−Φ ( z ) for all real z.

a−μ X−μ b−μ

P ( a ≤ X ≤b )=P( ≤ ≤ )
σ σ σ
a−μ b−μ
¿ P( ≤Z ≤ )
σ σ

¿Φ ( b−μ
σ )
−Φ (
σ )
a−μ

Example 9.2
A random variance X is normally distributed with mean 50 and standard deviation
10. Calculate P ( 45 ≤ X ≤62 ).
Solution
μ=50 ,σ =10 ⇒ X ∽ N ( 50 ,10 ).

45−50 X−50 62−50

P ( 45 ≤ X ≤62 )=P( ≤ ≤ )
10 10 10
¿ P(−0.5 ≤ Z ≤ 1.2)

¿ Φ ( 1.2 )−Φ (−0.5 )

 ¿ Φ ( 1.2 )−[1−Φ ( 0.5 ) ]

¿ 0.8849−[1−0.6915 ]=0.5764.

Example 9.3
In an examination the average mark was 76.5 and the standard deviation was 9.5.
If 15% of the class scored grade A and the marks are assumed to follow a normal
distribution, what is the lowest possible grade A mark and the highest possible
grade B mark?
Solution
X ∽ N ( 76.5 , 9.5 ).

Let a be the lowest possible grade A mark . Then

P ( X ≥ a )=0.15

(
a−76.5
Standardizing X we have P Z ≥ 9.5 =0.15 or )
P(Z ≤
9.5 )
a−76.5
=0.85 .

Using the normal tables we have

a−76.5
=1.04 ⇒ a=86.4
9.5

Therefore, the lowest grade A mark is 87, and the highest grade B mark, is 86.
Example 9.4
If a random variable X is normally distributed with mean μand variance μ2, and if
P ( X ≤ 8 )=0.95 , determine P ( 4 ≤ X ≤ 11 ).

Solution
X ∽ N (μ,μ)

(
P ( X ≤ 8 )=P Z ≤
8−μ
μ )=0.95

i.e Φ ( 8−μ
μ ) =0.95 ⇒
8−μ
μ
=1.65
 μ=3.02.

∴ P ( 4 ≤ X ≤11 )=P ( 4−μ

μ
≤Z ≤
μ )
11−μ

¿ P ( 0.32≤ Z ≤ 2.64 )

¿ Φ ( 2.64 )−Φ ( 0.32 )

¿ 0.9495−0.6255=0.3240.

Example 9.5 Let X be N ( μ , σ ) so that P ( X ≤ 89 )=0.90 and P ( X ≤ 94 ) 0.95. Find μ and σ 2.

Solution
P ( X ≤ 89 )=0.90 ⇒ P ¿

μ+1.28 σ =89 … … … … .. ( 1 )

Similarly
P ( X ≤ 94 ) =0.95 ⇒ P ¿

μ+1.65 σ =94 … … … ….. ( 2 )

Solving equations (1) and (2) simultaneously we have

μ=71.7 and σ 2=182.25

9.5 Assessment
1. Given that X is normal with mean 10 and variance 4, compute P (|X −10|>1.8 )
.
2. If X ∽ N ( 10 , σ ) and P ( X >12 )=0.1537, determine P ( 9< X < 11).

(X−μ
)
3. If X ∽ N ( μ , σ ), find the constant b so that P −b ≤ σ ≤ b =0.95 .
4. Let X be normally distributed with mean μand variance σ 2,and suppose that
( X ≤ 69 )=0.90 and P ( X ≤74 )=0.95 . Find μ and σ 2.
5. The time required to perform a certain job is a random variable having a
normal distribution with mean 50 minutes and a standard deviation of 10
minutes. Compute the probabilities that
 (i) the job will take more than 75 minutes,
(ii) the job will take less than 60 minutes,
(iii) the job will take between 45 and 60 minutes.

Gamma, Exponential and Beta Distributions

Gamma Distributions
A continuous random variable X is said to have the gamma distribution
with parameters α and β ( α >0 , β >0 ) if its p.d.f is of the form

{
β α α −1 −βx
f ( x )= Γ ( α ) x e ,∧x >0
0 ,∧otherwise

Therefore, we write X ∽ Γ ( α , β ) to mean that X has the gamma

distribution with parameters α and β .
Moments of the Gamma Random variable
The kth moment of a gamma random variable X about the point x=0 is given by
∞
E ( X )=∫ x f ( x ) dx
k k

α ∞
β
= ∫ x α +k−1 e− βx dx
Γ (α ) 0

β Γ ( α + k ) Γ ( α +k )
α
¿ . α + k =. k
Γ (α) β β Γ ( α)
 Γ ( α +1 ) α Γ ( α ) α
Thus E ( X )= ❑ = = which the mean of X.
β Γ (α ) β Γ ( α) β
Γ ( α +2 ) ( α + 1 ) αΓ ( α ) ( α +1 ) α
Now E ( X 2) = 2 = 2
= 2
β Γ (α ) β Γ (α ) β

2
Therefore Var ( X )=E ( X 2 )− [ E ( X ) ]
( α +1 ) α α 2 α
¿ 2
− 2= 2 .
β β β

The m.g.f of X is given by

M X ( t )=E ( e tX )
∞
¿ ∫ e f ( x ) dx
tx

x=0

α ∞
β
¿ ∫
Γ ( α ) x=0
tx α −1 − βx
e x e dx

α ∞
β
¿ ∫
Γ ( α ) x=0
α −1 −(β −t ) x
x e dx

Γ (α )
( )
α α
β β
¿ . = ,t <β
Γ ( α ) ( β−t )α β−t

We can now use this m.g.f of X to determine the mean and variance of X.
M X ( t )=α β α ( β−t )−α −1
∕

Therefore
α −α −1 α
E ( X )=M X ( 0 )=α β ( β )
∕
=
β

M X ( t )=α (α +1) β α ( β−t )−α −2

∕ ∕

α −α −2 ( α + 1) α
E ( X 2) =¿ M X ( 0 )=α ( α +1 ) β ( β ) =
∕ ∕
2
β
2
Therefore Var ( X )=E ( X 2 )− [ E ( X ) ]
 ( α +1 ) α α 2 α
¿ 2
− 2= 2.
β β β

Exponential Distribution
If a random variance X has density given by

{
−λx
f ( x )= λ e ,∧x >0 , λ> 0
0 ,∧otherwise

Then X is defined to have an exponential distribution.

∞
E( X )=∫ xf ( x ) dx
0

∞
1
¿ λ∫ x e
−λx
dx=
0 λ

∞
E( X )=∫ x f ( x ) dx
2 2

∞
2
¿ λ∫ x e
2 −λx
2 .
dx=
0 λ

2
Therefore Var ( X )=E ( X 2 )− [ E ( X ) ]
2 1 1
¿ 2
− 2= 2 .
λ λ λ

The moment generating function of an exponential random variable is given by

M X ( t )=E ( e tX )
∞
¿ ∫ e f ( x ) dx
tx

x=0

∞
¿λ ∫ etx e−λx dx
x=0

∞
λ
¿λ ∫ e−( λ−t )x dx= λ−t ,t < λ .
x=0
The mean and variance of X can be obtained by using the above m.g.f as follows:
∕ λ
M X ( t )=
( λ−t )2
∕ λ 1
∴ E ( X )=M X ( 0 )= =
( λ−0 ) λ
2

∕ ∕ 2λ
M X ( t )=
( λ−t )3
∕ ∕ 2λ 2
∴ E ( X ) =M X ( 0 ) =
2
3
= 2
( λ−0 ) λ
2
Therefore Var ( X )=E ( X 2 )− [ E ( X ) ]
2 1 1
¿ 2
− 2= 2.
λ λ λ

Example 10.1
Suppose that the number of minutes required to serve a costumer at service
counter has an exponential distribution with mean 2. Compute the probability
that the time required to serve a single costumer will exceed 4 minutes.
Solution
Let X denote the number of minutes required to serve a costumer at a service
counter. Then
1 1
E ( X )=2= ⇒ λ=
λ 2

Hence the p.d.f of X is given by

{
−1
1 2x
f ( x )= 2 e ,∧x >0
0 ,∧otherwise

[ ] =0.1353
∞ −1 −1 ∞
1 x x
∴ P ( X >4 )=∫ e 2
dx= −e 2
4
4 2

Beta Distribution
 A continuous random variable X is said to have the beta distribution
with parameters α and β ( α >0 , β >0 ) if its p.d.f is of the form

{
Γ ( α + β ) α −1 β−1
x ( 1−x ) ,∧0< x<1
f ( x )= Γ ( α ) Γ ( β )
0 ,∧otherwise

1
Γ (α ) Γ ( β )
NB: ∫ x α −1 (1−x )β −1 dx= Γ (α+β )
=Β ( α , β )
0

Moments of Beta Random variable

The moments of the beta distribution are computed as follows:
1
E ( X )=∫ x f ( x ) dx , k =1 ,2 , …
k k

1
Γ ( α+β)
¿ ∫
Γ (α) Γ ( β ) 0
k α−1 β −1
x x (1−x ) dx

1
Γ ( α+β)
¿ ∫
Γ (α) Γ ( β ) 0
❑x
α +k−1 β −1
( 1−x ) dx

Γ ( α + β ) Γ ( α +k ) Γ ( β )
¿ .
Γ ( α ) Γ ( β ) Γ ( α + β+ k )

Γ ( α + β ) Γ ( α +k )
¿ .
Γ ( α ) Γ ( α + β+ k )
Putting k=1 we have
Γ ( α + β ) Γ ( α + 1)
E ( X )= .
Γ ( α ) Γ ( α + β +1 )

Γ (α+β ) αΓ ( α ) α
¿ . = .
Γ (α ) (α+β ) Γ (α+ β ) α +β

For k=2 we have

Γ ( α + β ) Γ ( α +2 )
E ( X 2) = .
Γ ( α ) Γ ( α + β +2 )
 Γ (α+β ) α ( α +1 ) Γ ( α )
¿ .
Γ ( α ) ( α + β ) ( α + β+1 ) Γ ( α + β )

α (α + 1)
¿
( α + β ) ( α + β+1 )

2
Therefore Var ( X )=E ( X 2 )− [ E ( X ) ]

α (α + 1) α
2
¿ −
( α + β ) ( α + β+1 ) ( α + β )2

αβ
¿ 2 .
( α + β ) ( α + β +1 )

10.3 : Assessment
1. Show that, if in gamma density α =1, then the gamma density specializes
exponential density.
2. Show that the beta distribution reduces to the uniform distribution over
(0,1) if α =β=1.

FUNCTIONS OF A RANDOM VARIABLE

Change of variable
Variables with Discrete Distributions
Suppose that a random variable X has a discrete distribution for which the
probability function is f ( x ) . Let U =Φ ( X ) be another random variable defined as a
function of X.
∴ g (u )=P ( U=u )

¿ P ¿ Φ ( X )=u ¿

¿ ∑ f (x )
x: ϕ ( x ) =u

Example 11.1
Let X have the binomial distribution given by

{( )( ) ( )
x
4 3 1 4− x
,∧x=0 , 1 ,2 , 3 , 4
f ( x )= x 4 4
0 ,∧otherwise

Find the distribution of U =X 2.

Solution
The possible values of U are 0,1,4,9,16
∴ g (u )=P ( U=u )

¿ P ( X 2=u )=P ( X =√ u )

{( )( ) ( )
√u
4 3 1 4−√u
,∧u=0 , 1 , 4 , 9 , 16
g ( u )= √ u 4 4
0 ,∧otherwise

Example 11.2
Suppose that X has the discrete distribution given in the following table:
x -3 -2 -1 0 1 2 3
f (x) 4 1 1 1 1 1 4
21 6 7 1414 6 21
Find the distribution of the random variable U =3 X 2 +1.
Solution
The possible values of U are 1,4,13,28.
1
P ( U=1 ) =P ( X=0 )=
7
1 1 1
P ( U=4 ) =P ( X=−1∨X=1 )=P ( X=−1 ) + P ( X =1 )= + =
14 14 7
1 1 1
P ( U=13 )=P ( X=−2∨X =2 )=P ( X=−2 )+ P ( X=2 ) ¿ + =
6 6 3
P ( U=28 )=P ( X=−3∨X =3 )=P ( X =−3 ) + P( X=3)
4 4 8
¿ + = .
21 21 21

Hence the probability distribution of U is

u 1 4 13 28
g(u) 1 1 1 8
7 7 3 21

11.3.2: Variables with Continuous Distributions

Suppose X is a random variable with p.d.f f (x) . Let U =Φ ( X ) be another random
variable. Then for any real number u, the cumulative distribution function G(u) of
U is
G ( u )=P(U ≤ u)

¿ P(Φ ( X ) ≤ u)
❑
¿ ∫ f ¿ ¿x) dx
x: ϕ (x)≤u

If G(u) is a continuous function of u, then at any point u at which G is

differentiable the p.d.f of u will be given by
dG(u)
g ( u )=
du

Example 11.3 Let X have a p.d.f given by

 {
f ( x )= 2 x ,∧0< x <1
0 ,∧otherwise

Find the p.d.f of U =2 X +3.

Solution
The cumulative distribution of U is given by
G ( u )=P(U ≤ u)

¿ P(2 X +3 ≤ u)

¿ P¿
1
(u−3)
2

¿ ∫ f ( x ) dx
0

1
(u−3)
2

¿ ∫ 2 x dx
0

1
(u−3)
1
¿ [ x 2 ]0
2 2
= ( u−3 )
4

dG(u) 1
∴ g (u )= = ( u−3 )
du 2

The range of u is 3<u< 5. Hence

{
1
g ( u )= 2
( u−3 ) ,∧3<u<5
0 ,∧otherwise

Example 11.4
Suppose that X has a uniform distribution on the interval (−1 , 1). Find the p.d.f of
U =−ln |X|.
Solution
In this case the p.d.f of X is
 {
1
,∧−1< x <1
f ( x )= 2
0 ,∧otherwise

The cd.f of U is given by

G ( u )=P(U ≤ u)

¿ P(−ln| X|≤u)

¿ P(| X|≥ e )
−u

¿ 1−P (| X|≤ e−u )

¿ 1−P (−e−u ≤ X ≤ e−u )

−u
e
¿ 1− ∫ f ( x ) dx
−u
−e

−u
e
1
¿ 1− ∫ dx=1−e−u
−e
−u 2

Therefore, the p.d.f of U is

dG(u)
g ( u )=
du

{
−u
g ( u )= e ,∧u> 0
0 ,∧otherwise

Direct derivation of the density function for a Continuous Random Variable

Suppose X is a random variable and U =Φ ( X ) . If a< x< b andα <u< β . Let x=ω (u ), then
the function ω is the inverse of Φ.
If we assumed that the function Φis continuous and strictly increasing over the
interval ( a , b ) ,the inverse function ω is also continuous and strictly increasing over
the interval ( α , β ). Hence for any value u such that α <u< β ,
G ( u )=P(U ≤ u)

¿ P(Φ ( X ) ≤ u)
 ¿ P(X ≤ ω ( u ))

¿ F ( ω ( u) )

Where F is the distribution function of X.

If we now assume in addition that ω is a differentiable function over the interval
( α , β ) , then the distribution of U is continuous and its p.d.f is given by

dG(u)
g ( u )=
du
dF ( ω ( u ) )
¿
du
dω ( u )
¿ f ( ω (u) ) for ¿ u< β .
du

Similarly if Φis continuous and strictly decreasing over the interval ( a , b ), then U
will vary over some interval ( α , β ) as X varies over the interval ( a , b ), and the
inverse function ωwill be continuous and strictly decreasing over the interval ( α , β )
. Hence for α <u< β ,
G ( u )=P(U ≤ u)

¿ P(Φ ( X ) ≤ u)

¿ P(X ≥ ω ( u ))

¿ 1−P( X ≤ω (u ))

¿ 1−F ( ω ( u ) )

If ω is differentiable over the interval ( α , β ), then

dG(u)
g ( u )=
du
d [1−F ( ω (u ) ) ]
¿
du
dω ( u )
¿−f ( ω (u ) ) for α <u< β .
du
 dω ( u )
Since ω is strictly decreasing, <0 and hence g ( u )can be expressed in the
du
form

g ( u )=f ( ω ( u ) ) | |.
dω ( u )
du

Example 11.5
Let X be a random variable with p.d.f given by

{
1
x ,∧0< x< 2
f ( x )= 2
0 ,∧otherwise

Find the p.d.f of a new random variable U =1−X 2.

Solution
−3<u< 1

u=1−x ⇒ x =√ 1−u
2

dx −1
=
du 2 √1−u

∴ g ( u )=f ( ω ( u ) ) | |
dω ( u )
du

¿ f (x) |dxdu|
1
¿ f ( √ 1−u )
2 √ 1−u
1 1 1
¿ √ 1−u . =
2 2 √1−u 4

{
1
,∧−3 <u<1
¿ 4
0 ,∧otherwise