L4: Special continuous distribution

Uniform distribution (rectangular)

Let the r.v. X denote the outcome when a point is
selected at random from an interval
[a, b],   a  b   .

If the experiment is performed in a fair manner, it

is reasonable to assume that
( x  a)
P(the point is selected from the interval [a, x], a  x  b ) =
(b  a )
That is, the probability is proportional to the length
of the interval so that the d.f. of X is

 0 ; xa
 x  a
F ( x)   ; a xb
b  a
 1 ; bx

Because X is a continuous-type r.v., F ( x) is equal

to the p.d.f. of X whenever F ( x) exists. Thus
1
when a  x  b , we have F ( x)  .
ba

The r.v. X has a uniform distribution if its p.d.f. is

 1
 ; a xb
f ( x)   b  a
 0 ; otherwise

1
Notation: X ~ U (a, b)

Also called rectangular distribution because the

graph of f ( x) suggests the name.

2
Using the same method, it can be shown that
etb  eta
m(t )  ; t0
t (b  a )

Example: Every morning, a student is waiting for a

bus to go to classes. The bus will arrive at 7.10 am
and every 15 minutes after that. Let say the time
that the student arrives at the bus stop is uniformly
distributed between 7.00 and 8.00 am. What is the
probability that he has to wait at least 5 minutes?

3
Normal distribution

-the most important distribution in statistical

applications since many measurements have
(approximate) normal distribution.

The r.v. X has a normal distribution if its p.d.f. is

defined by
1  ( x   )2 
f ( x)  exp    ,   x  
 2  2 
2

where  and  are parameters satisfying

    ,0     and exp[v] means ev .

Notation: X ~ N (  , 2 )

Refer to Handout 4a and 4b.

4
To find the mean and the variance, we will first
find the m.g.f..

Refer to Handout 4c and 4d.

5
Example:
1  ( x  7) 2 
f ( x)  exp    ,   x  
32  32 

If Z ~ N (0,1), we shall say that Z has a standard

normal distribution. The d.f. of Z is
z
 ( z )  P (  Z  z )   e w
2
1 /2
2
dw


The values can be obtained from the table - Table

of Normal distribution

6
The graph shown is the graph of the p.d.f. of Z .
The shaded area =  ( z0 ) . Because of symmetry,
 ( z )  1   ( z ) for z  R .

Example: If Z ~ N (0,1),
a) compute (i) P (1.25  Z  2.75)
(ii) P (1.65  Z  0.70)

b) find the constants a, b and c such that

(i) P (0  Z  a )  0.4147
(ii) P ( Z  b)  0.05
(iii) P(| Z | c)  0.95

7
Theorem: If X is N (  , 2 ) , then
X 
Z is N (0,1) .


This theorem can be used to find probabilities

about X , which is N (  , 2 ) as follows:
a X  b 
P ( a  X  b)  P   
    
b  a 
      
     
X 
since is N (0,1).


8
Example: X ~ N (25,36) . Find c such that
P (| X  25 | c)  0.9544 .

9
Exponential distribution

W  waiting time until the first change occurs

when observing a Poisson process in which the
mean number of changes in the unit interval is  .

Then W is a continuous-type r.v.. Clearly, this

waiting time is nonnegative. Thus

We say that the r.v. X has an exponential

distribution if its p.d.f. is defines by

1
f ( x)  e x / , 0  x  


where the parameter   0 . Accordingly, the

waiting time W has an exponential distribution
with   1/  .

10
To determine the exact meaning of the parameter
 , we first find the m.g.f..

11
So if  is the mean number of changes in the unit
interval, then   1/  is the mean waiting time for
the first change. In particular, suppose that   7 is
the mean number of changes per minute, then the
mean waiting time for the first change is 1/ 7 of a
minute.

Example: Let X have an exponential distribution

with a mean of 100.

12
Let X have an exponential distribution with mean
   . Then the distribution function of X (as
we’d worked out) is

 0,   x  0
F ( x)    x /
1  e , 0  x  

It is useful to note that for an exponential r.v., X ,

we have that
P ( X  x)  1  F ( x)  1  (1  e  x /  )  e  x / 

Example: Customers arrive in a certain shop

according to an approximate Poisson process at a
mean rate of 20 per hour. What is the probability
that the shopkeeper will have to wait more than 5
minutes for the arrival of the first customer?

13
Example: Suppose that the life of a certain type of
electron has an exponential distribution with a
mean life of 500 hours. If X denotes the life of a
tube (or the time to failure of a tube), then

P( X  x)   1
500 e  t / 500 dt  e  x / 500
x

Suppose that the tube has been in operation for 300

hours. The conditional probability that it will last
for another 600 hours is

14
 Gamma distribution

In the (approximate) Poisson process with mean  ,

let W denote the waiting time until the h th change
occurs.

15
However,

(1)   e  y dy  1.
0
Thus, when n is a positive integer, we have that
(n)  (n  1)!
And for this reason, the gamma function is called
the generalized factorial. Incidentally, (1)
corresponds to 0!, and we have noted that (1)  1,
which is consistent with earlier definitions.

16
The r.v. X has a gamma distribution if its p.d.f. is
1
f ( x)  
x 1e  x /  , 0  x  
( )

Hence, W , the waiting time until the h th change in

Poisson process, has a gamma distribution with
parameter   h and   1/ 

Refer to Handout 4e.

17
The m.g.f. of X is
1 1
M (t )  , t
(1   t ) 

The mean and variance are

   and  2   2

Example: Suppose that an average of 30 customers

per hour arrive at a shop in accordance with a
Poisson process. What is the probability that the
shopkeeper will wait more than 5 minutes before
both of the first two customers arrive?

18
From F ( x) with   1/  (since   h is an
integer),

( x /  )k e x /
 1
P( X  x)   .
k 0 k!

Thus, with x  5,  2, and   2 , this is equal to

19
Another special case of Gamma distribution is a
Chi-Square distribution

Let X have a gamma distribution with   2 and

  r / 2 , where r is a positive integer. The p.d.f. of
X is

1 ( r / 2) 1  x / 2
f ( x)  x e , 0  x  .
(r / 2) 2 r/2

We say that X has a chi-square distribution with

r degree of freedom.

Notation: X ~  2 (r )

The mean and variance are

    r  (d .k .) and  2   2  2r  (2 d .k .) .

Its m.g.f. is M (t )  (1  2t )  r / 2 , t  12 .
Refer to Handout 4f.

20
Example: If customers arrive at a shop on the
average of 30 per hour in accordance with a
Poisson process, what is the probability that the
shopkeeper will have to wait longer that 9.390
minutes for the first nine customers to arrive?

Example: If X has an exponential distribution with

a mean of 2, the p.d.f. of X is

21
 Beta distribution

A beta function is defined by:

1
B ( ,  )   x 1 (1  x)  1dx,   0,   0
0

The relationship between the beta function and the

gamma function:
( )(  )
B ( ,  )  ,
(   )
which gives us an alternative (and simpler) way to
calculate B( ,  ) .

The r.v. X has a beta distribution if its p.d.f. is

1
f ( x)  x 1 (1  x)  1 , 0  x  1
B ( ,  )
(   )  1
 x (1  x)  1 , 0  x  1,  0,   0
( )(  )

It is clear that B ( ,  ) is the constant that makes f

a density.

Notation: X ~ Beta( ,  )

When     1, Beta ( ,  )  U (0,1) .

Refer to Handout 4g.

22
The beta distribution is frequently used to model
uncertain fractions. For example, the fraction of
defective items when sampling with replacement
and we have no prior opinion about the fraction of
defects in the entire lot of manufactured items.

It is often used as a model for proportions, such as

the proportion of impurities in a chemical product
or the proportion of time that a machine is in a state
of being repaired.

The beta density can be derived from the Binomial

distributions using Bayes’s theorem.

If k is a positive integer, the k th moment is

E( X k ) .

23
The cumulative distribution function for the beta
r.v. is commonly called the Incomplete beta
function and is denoted by

t  1 (1  t )  1
x
F ( x)   dt  I x ( ,  )
0 B ( ,  )
where the values is given in Tables of the
Incomplete Beta Function.

Next example involves the case when  and  are

both integers.

Example: A gasoline wholesale distributor has

bulk storage tanks that hold fixed supplies and are
filled every Monday. Of interest to the wholesaler
is the proportion of this supply that is sold during
the week. Over many weeks of observation it was
found that this proportion could be modeled by a
beta distribution with   4 and   2 . Find the
probability that the wholesaler will sell at least
90% of her stock in a given week.

24