Hypergeometric Distribution

Chucky Chung

1
 1. PMF

Let N ∈ N and K, n ∈ {0, 1, ..., N }

(K )(N −K )
f (x) = x
) ,
(Nn−x x ∈ {max(0, n + K − N ), ..., min(n, K)}
n

(Remark) Although the support {max(0, n + K − N ), ..., min(n, K)} is valid, it

is inconvenient to calculate or derived the properties. One usual way

to fix this problem is to change the support to {0, ..., n}, where the set

make sure that all of the numbers in combinations are non-negative.

This two supports are identical if 0 ≥ n + K − N and n ≤ k. For

(n)
the rest of situations, since we have k
= 0 whenever k > n, which
−K
(Kx )(Nn−x )
also implies that = 0. Thus we can changed the support to
(n)
N

{0, ..., n} with out the value of pmf changes.

2. Sum to 1.

∑
f (x)
x∈Sx

(K )(N −K )
∑
n
= x
(Nn−x
)
x=0 n

n ( )( )
1 ∑ K N −K
= (N )
n x=0
x n−x

( )
1N
= (N ) =1 (See Appendix A.)
n
n

2
3. Expected value

For r > n, E[(X)r ] = 0. Now consider n ≥ r.

E[(X)r ]

∑
= (x)r f (x)
x∈Sx

(K )(N −K )
∑
n
= x
(Nn−x
) (x)r
x=0 n

(K )(N −K )
∑
n
= x
(Nn−x
) (x)r
x=r n

n ( )( )
1 ∑ K N −K x!
= (N )
n x=r
x n − x (x − r)!

If K < r, the sum is equals to 0.

Now we consider K ≥ r

E[(X)r ]

( −K )
1 ∑
n
K! Nn−x
= (N )
n x=r
(K − x)!(x − r)!

(N −K )
(K)r ∑ (K − r)! n−x
n
= (N )
n x=r
(K − x)!(x − r)!

n ( )( )
(K)r ∑ K − r N − K
= (N )
n x=r
x−r n−x

3
 n−r ( )( )
(K)r ∑ K − r N −K
= (N )
n y=0
y n−r−y

( )
(K)r N − r (K)r (n)r
= (N ) = for (N )r ̸= 0
n
n − r (N ) r

(K)r (n)r
Since (N )r
= 0 when K < r or n < r. This equation holds for all

such n, K.

(K)r (n)r Kn
E[X] = = .
(N )r N

4. Variance

V ar(X)

= E[(X)2 ] + E[X] − E[X]

(K)2 (n)2
= + E[X](1 − E[X])
(N )2

K(K − 1)n(n − 1)
= + E[X](1 − E[X])
N (N − 1)

(K − 1)(n − 1)
= E[X]( ) + E[X](1 − E[X])
(N − 1)

(K − 1)(n − 1)
= E[X](1 − E[X] + )
(N − 1)

N (N − 1) nK(N − 1) N (K − 1)(n − 1)
= E[X]( − + )
N (N − 1) N (N − 1) N (N − 1)

4
 (N − nK)(N − 1) + N (K − 1)(n − 1)
= E[X]( )
N (N − 1)

nK + N 2 − N K − N n
= E[X]( )
N (N − 1)

nK (N − n)(N − k)
=
N N (N − 1)

nK (N − K) N − n
=
N N N −1

5. MGF MGF exists but too complicated.

5
Appendix A.

( ) ∑k ( )( )
n+m m n
Show that =
k r=0
r k−r

For the cases k = 0, both sides equal to zero.

Now we consider k > 0(i.e. k − 1 ≥ 0), and used induction on n.

For n = 0, notice that

( ) ( ) ( )
0+m m 0
= and = 0 whenever k − r > 0.
k k k−r

( ) ( ) ∑k−1 ( )( ) ∑ k ( )( )
0+m m m 0 m 0
= + = .
k k r=0
r k − r r=0
r k − r

This gives us the basic case.

Now suppose inductively that n = p is true.

For n = p + 1

( )
m + (p + 1)
k

( ) ( )
m+p m+p
= +
k k−1

∑k ( )( ) ∑k−1 ( )( )
m p m p
= +
r=0
r k−r r=0
r k−1−r

( ) ∑k−1 ( )( ) ∑k−1 ( )( )
m m p m p
= + +
k r=0
r k − r r=0
r k−1−r

6
 ( ) ∑k−1 ( ) ( ) ( )
m m p p
= + [ + ]
k r=0
r k − r k − 1 − r

( )( ) ∑k−1 ( )( ) ∑ k ( )( )
m n+1 m p+1 m p+1
= + =
k 0 r=0
r k − r r=0
r k−r

(Remark) This equation also known as Vandermonde’s identity.