PROBABILITY DISTRIBUTIONS

Hyper-geometric Distribution
In drawing a random sample with replacement, one object is drawn from the sample
space and put it back to the given set, mixing the latter before the next draw. This guarantees
independence of trials and leads to the binomial distribution.
Suppose that a sample of size n is to be chosen randomly from the population of N items, of
which M are of a special type (say defective) and remaining ( N  M ) are of other type (say
serviceable items). Let X denotes the number of special items out of selected sample of size n,
M
then the probability of drawing a defective item in a trial is p  . Hence in drawing a sample
N
of n items, with replacement, the probability of obtaining r defective items is
r nr
M   M 
P ( X  r )  nCr    1   , r  0, 1, 2,  , n .
N   N 

In sampling without replacement, in which no items are replace back to the box, the
M
Cr N  M C n  r
probability is P( X  r )  N
, r  0, 1, 2,  , n . This distribution is known as the
Cn
Hyper-geometric distribution. We write X  H ( n, N , M ) .

Note: Although the value of r should be between 0 and n, it must satisfy

n  ( N  M )  r  min( n, M ) .
nM nM ( N  M )( N  n)
The distribution has mean E ( X )  and the variance Var ( X )  .
N N 2 ( N  1)

Note 1: If N, M and ( N  M ) are large compared with sample size n, then it does not matter
too much whether the sample is with or without replacement, and the Hyper-geometric
distribution may be approximated by the binomial distribution.
( N  n)
Note 2: The factor is known as Finite Population Correction Factor.
( N  1)

Exponential Distribution
A continuous random variable X assuming non-negative values is said to have Exponential
distribution if its p.d.f. reduces exponentially from   0 at X  0 to the value zero
asymptotically as X tends to  . The p.d.f. is given by f ( x)   e   x , x  0.
For example, Life of an electric components, Radioactive decay, Period between two
successive occurrence of an event, Time required for repairs, Time require to find a fault.

Cumulative Density function (c. d. f.) FX ( a )  P ( X  a )  1  e  a .

Note:
1) P ( X  a )  e a
1
2) Mean of the distribution is E ( X )    and

Page 1 of 3
 1 1
3) The variance is Var ( X )  , so Standard deviation s.d .  .
 2


Memory-less or Forgetfulness Property: Let X be an exponential random variable with mean

1 P ( X  s  t ) e   ( s t )
  then P ( X  s  t / X  s)     s  e  t  P ( X  t ), s , t  0.
 P( X  s) e
This is the memory-less property of the distribution. Exponential distribution is the only
continuous distribution with memory-less property.

Note: Exponential distribution is related to Poisson distribution in some sense. The number of
events, which occur in [0, t ] has Poisson distribution. The time that elapses between such
events follows exponential distribution.
Relationship between Exponential and Poisson distributions:

e   t ( t ) x
If X is a Poisson random variable, then P ( X  x )   .
x 0 x!
Probability of no event to occur upto time t is P ( X  0)  e  t .
The probability that the length of time until the first event will exceed t is same as the
probability that no Poisson events occur in the period t. If X is the random variable representing
the time to the first Poisson event then,
P( no event occur in time t) = e t  P ( X  t )
Therefore the c.d.f. of X is FX ( x)  P (0  X  x)  1  e   x
Hence the density function is f X ( x)   e   x , which is nothing but the density function of
1
exponential distribution with    .

Gamma Distribution
A random variable X is said to follow Gamma distribution with parameters  ,  if its
probability density function p.d.f. is
x 1e  x /  
f ( x)  
, x  0,  ,   0 where ( )   x e dx .
 1  x

  ( ) 0

If X is Gamma distributed with parameters  ,  then it is denoted as X   ( ,  ) .

Note:

1. The distribution has the mean  and the variance  2 .

2. It is a skewed distribution.
1   x 1e   x
3. Putting   the p.d.f. reduces to f ( x )  , x  0.
  ( )
4.  is called the shape parameter and  is called the rate parameter (it is the not at
1
which the events occur).   is called the scale parameter.

5. If  is a positive integer Gamma distribution is known as Erlang distribution.
Page 2 of 3
6. If   1 , Gamma distribution is known as Exponential distribution.
7. Putting y   x , the probability density function of the random variable Y
y 1e  y
f ( y)  , y  0.
( )
It is known as the standard Gamma distribution with parameter  , it is independent
of  .
8. For large  , (  50) , standard Gamma function resembles a Normal distribution
with mean and variance approximately equal to  .
 1
9. The Gamma distribution with parameters   and   reduces to the Chi-square
2 2
distribution with  degrees of freedom.
10. If a system consisting of one original components and (r  1) spare components such
that when the original component fails, one of the (r  1) spare components is used. If
this component fails, one of the (r  2) spare components is used. System fails only
when the original component and all the (r  1) spare components fail. If the lifetime
T1 , T2 ,..., Tr , of all the duplicate components have the same exponential distribution
with parameter  and are probabilistically independent, then the lifetime (time until
r
failure) of the entire system is the sum X   Ti having Gamma distribution with
i 1

 x e
r r 1   x

density function f ( x )  , x  0. .
(r )

Page 3 of 3