3.

2 SPECIAL UNIVARIATE CONTINUOUS PROBABILITY DISTRIBUTIONS

3.2.1 The uniform distribution

Consider the figure below.

1 f(x)

a b

We can determine the area under the curve in two ways:

(a) Area= (b  a) 1 sq. units

b b b
(b) Area=  f ( x)dx   1 dx   dx
a a a

Therefore
b
b  a   dx ………………………………………………………………………………(i)
a

Divide (i) by (b-a) to obtain

b 1
1  dx ……………………………………………………………………………..(ii)
a b 1

Because an integration of a p.d.f of a random variable over its entire support

yields 1, the integrand of (ii) must be a valid p.d.f.

Therefore

 1
 , a xb
f ( x)   b  a is called the probability distribution of a uniformly

 0, otherwise
distributed random variable.

1
Some distributional quantiles

Suppose X is a random variable with a uniform distribution in the interval a<x<b.

The mean and variance are respectively

1 1
  a  b  and  2   b  a 
2

2 12

Proof
b
E ( X )   xf ( x)dx
a
b 1
 x dx
a ba
1 b
b  a a
 xdx
b
1  x2 
  
b  a  2 a
1 (b  a) 2

ba 2
ba

2
b
E ( X 2 )   x 2 f ( x)dx
a
b 1
  x2 dx
a ba
1 b 2
b  a a
 x dx
b
1  x3 

b  a  3  a
1 (b  a)3

ba 3
1 b  3ab 2  3a 2b  a 3
3

ba 3
b  2ab  a
2 2

3

Hence

2
Var ( X )   2  E ( X 2 )   E ( X ) 
2

b 2  2ab  a 2  b  a 
2

  
3  2 
1
 b  a 
2

12

3.2.2 The Gamma Distribution

Consider the gamma function of  defined by

( )   y 1e  y dy ………………………………………………………………………(1)
0

Divide both sides of (1) by ( ) to obtain

 y 1e y dy
1  …………………………………………………………………………..(2)
0 ( )

Since the integration of a p.d.f over its entire support equals to 1, we conclude
that the integrand of (2) is a valid p.d.f.

 x 1e x
 , x  0,   0
Hence f ( x)   ( ) …………………………………………………..(3)

 0. elsewhere

Equation (3) is called a one parameter gamma probability distribution.

Y
Furthermore, if we consider the transformations X   Y and X  , we can

derive respectively type I and type II gamma distributions with two parameters
 and  .

y dy
To derive the latter, let x     . Substituting this result in (3) gives
 dx
 1
 y
y


  e
g ( y)    
( )
y

 1 
y e
 
 ( )

3
  1  1 

x

 x e , x  0, ( ,   0)
Therefore f ( x)     ( ) is the Type II gamma distribution

 0, otherwise
with parameters  and  .

Note


( )   y 1e y dy , α > 0
0

=   1   1

=   1  2    2 

  1  2   3   3
=.
.
.

=   1  2   3 ...1

=   1!

For this reason,  is sometimes called the factorial notation. An important

property for the gamma function is that

 1
( )(1   )  so that if α = ½ we obtain     
sin  2

Some distributional Quantiles

4
The mean and variance of the gamma distribution are respectively given by

1
μ = αβ, σ2 = αβ2. And the moment generating function is m(t )  .
1   t 


Proof


M (t )  E (etx )   etx f ( x)dx
0
 1 x
  etx 
x 1e  dx
0  ( )
x
1  tx 


 1 
 
x e dx
 ( ) 0

 1  t 
1   x

 ( ) 0
 1   
 
x e dx

 1 t  u  du
Let u   x x  dx 
   1 t 1 t

Thus

 1
1   u   du
M (t )   
 ( ) 0  1   t 
eu
1 t
1 
 0
 u 1e  u du
( )(1   t )
1
 ( )
( )(1   t )
1

(1   t )

Next we obtain the first and second moments about the origin.

5
 d
E( X )  M (t ) |t 0
dt
d 1
 |t 0
dt (1   t )
d
 (1   t )  |t 0
dt
 ( )(1   t )  1 (  ) |t 0
 

which is the mean a gamma distributed random variable.

d2
E ( X )  2 M (t ) |t 0
2

dt
d
  (1   t )  1 |t 0
dt
  (  1)(1   t )   2 (  ) |t 0
  2 (  1)

Hence

Var ( X )  E ( X 2 )   E ( X ) 
2

  2 (  1)  ( ) 2
  2  2   2   2  2
  2

Note

The first and second moments about the origin may also be obtained from the r-

th moment about the origin or by using first principles of mathematical

expectation.

6
Example

A random variable X is gamma distributed with α = 2 , β = 3. Find P(x≤1).

Solution.

Since X is gamma distributed, the density function is

 x 1e 
x

 , x>0
f ( x)     
0,
 otherwise

We are given α = 2 , β = 3. Therefore

x 21e 3
x

f ( x)  2
3 2

 xe  3
x

 , x>0
= 9 2
0,
 elsewhere

But 2 =(2-1)! = 1! = 1.

 xe  3
x

 , x>0
Thus f ( x)   9
0,
 elsewhere

xe 3
1 x

a. P  x  1   dx
0
9

1
1
=  xe 3 dx
x

90

7
 dv du
Recall  u dx dx  uv   v dx dx, which is the formula for integration by parts.

Therefore

1
   
1
p( x  1)   x 3e 3   3e 3 .1dx 
x x

9 0

1 1
=  3 xe 3  9e 3 
x x

9 0

1
 1 x x 
=   xe 3  e 3 
 3 0

 1 1 1 
=   e 3  e 3    0  e0 
 3 

4 1 4 1
=  e 3  1  1  e 3 =0.0446
3 3

x
2 3
xe
b. p(1  x  2)   dx
1
9

2
1 x
  xe dx
3

91

   
2
1 x x 
=  x 3e   3e .1dx 
3 3

9 1

1 2
=  3 xe 3  9e 3 
x x

9 1

2
 1 x x 
=   xe 3  e 3 
 3 1

8
  1 2 2   1 1 1 
= e 3  e 3   e 3  e 3 
 3   3 

4 1 5 2
=  e 3  e 3 = .0998
3 3

3.2.3 The Exponential Distribution

This is one of the special cases of the gamma density with α = 1. Therefore a

random variable X is said to be exponentially distributed or has an exponential

distribution if the density function is

 e  x
 , x>0
f ( x)   
0,
 elsewhere

The mean and variance of an exponential distribution are given by    , and

1
 2   2 Its moment generating function is m(t ) 
1   t 