STAT 2001 Tutorial 3

STAT 2001 Basic Concepts in Statistics and Probability

I
Tutorial 3
September 20 – October 4, 2024
In this tutorial note, we will cover the following topics

• Expectation and variance of discrete random variable

• Binomial distribution

• Hypergeometric Distribution

• Geometric Distribution

• Negative Binomial Distribution

1 Expectation and variance of discrete random variable

Recall some definition from tutorial note 2. Let X be a random variable , and let S be the
sample space of X.

Definition 1. (Discrete random variable)

A random variable is called discrete if it has either a finite or a countable number of possible
values.

Definition 2. (pmf)
The Probability Mass Function (p.m.f.) of a discrete random variable X is a function
f (x) satisfies

1. f (x) > 0, ∀x ∈ S;
P
2. x∈S f (x) = 1.

3. p(x ∈ A) = f (x), ∀A ⊂ S;
P
x∈A

Definition 3. (Expectation)
P
• E(X) = x∈S xf (x)
P
• E(u(X)) = x∈S u(x)f (x), where u is a real-value function over S.

Definition 4. (Variance and standard deviation)

• V ar(X) = E(X − E(X))2

q
• σ= V ar(X)

Properties (E(X), V ar(X))

1. V ar(X) = E(X − E(X))2 = E(X 2 ) − 2E(X)E(X) + [E(X)]2 = E(X 2 ) − [E(X)]2

2. E(c1 X1 + c2 X2 ) = c1 E(X1 ) + c2 E(X2 ), where c1 , c2 are constants.

1
 STAT 2001 Tutorial 3

3. E (c1 u1 (X1 ) + c2 u1 (X2 )) = c1 E (u1 (X1 )) + c2 E (u2 (X2 )), where u1 , u2 are a real-value
function over S.

4. V ar(cX) = c2 V ar(X), where c is constants.

5. V ar (cu(X)) = c2 V ar (u(X)), where u is a real-value function over S.

6. V ar (c1 u(X) + c2 ) = c21 V ar (u(X)), where u is a real-value function over S.

Exercise 1. An insurance company sells an automobile policy with a deducible of one unit.
Let X be the amount of the loss having p.m.f.

0.9 , x=0
f (x) = c
 , x = 1, 2, ..., 6,
x
where c is a constant. Determine c and the expected value, variance of the amount the
insurance company must pay.

2
 STAT 2001 Tutorial 3

Exercise 2. Let X be a random variable with support 1, 2, 3, 4, 5, hand eachipoint has equal
probability. Argue that E(X) is the value that minimizes h(c) = E (X − c)2 . If the support
of X or the p.m.f. of X is changed, does E(X) still has this optimal property?

2 Binomial distribution
Binomial Distribution describes the behavior of n Bernoulli trials, each having a probability of
success p.
Summary
For X ∼ Bin(n, p),
!
n x
1. The p.m.f: f (x) = p (1 − p)n−x ,x = 0, 1, 2, ..., n.
x
2. E(X) = np, V ar(X) = np(1 − p)

3
 STAT 2001 Tutorial 3

Exercise 3. Suppose that the percentage of American drivers who are multitaskers(e.g., talk
on the cell phone, eat a snack) is approximately 65. In a random sample of n = 20 drivers, let
X equal the number of multitaskers.

1. How is X distributed?

2. Given the values of the mean, variance and standard derivation of X.

3. Find P (X = 15) and P (X ≥ 2).

Exercise 4. Consider a binomial random variable X with parameter n and p. Let m be the
largest integer that is less than or equal to (n+1)p. Show that the p.m.f. of X is monotonically
increasing with x for 0 ≤ x < m and is monotonically decreasing with x for x ≥ m. Then m
is the mode of X, which is defined to be the point of X that has the maximum probability.

4
 STAT 2001 Tutorial 3

3 Hypergeometric Distribution
The Hypergeometric Distribution describes the probability of x successes in n draws, without
replacement, from a finite population of size N that contains exactly K successes objects.
Summary
For X ∼ Hypergeometric(N, K, n),
! !
K N −K
x n−x
1. The p.m.f: f (x) = ! ,x ≤ n, x ≤ K and n − x ≤ N − K.
N
n
K N −n
 
2. Let = p, E(X) = np, V ar(X) = np(1 − p)
N N −1
Remark 1. Binomial distribution describes the probability of k successes in n draws with
replacement.

Remark 2. When N is sufficiently large, Hypergeometric distribution can be approximate by

Binomial Distribution.

4 Geometric Distribution
The geometric distribution represents the number of failures before you get a success in a series
of Bernoulli trials, each having a probability of success p.
Summary
For X ∼ Geom(p),

1. The p.m.f: f (x) = (1 − p)x−1 p,x = 1, 2, ..., n.

1 1−p
2. E(X) = , V ar(X) =
p p2
Property of Loss of Memory

P (X > k + j|X > k) = P (X > j),

where k and j are non-negative integers.
That means if you intend to repeat an experiment until the first ’success’, then, given that
the first ’success’ has not yet occurred, the conditional probability distribution of the number
of additional trials does not depend on how many failures have been observed. Geometric
Distribution is the only discrete distribution which has this property.

5
 STAT 2001 Tutorial 3

Exercise 5. Find the c.d.f of Geometric Distribution

Exercise 6. Prove the property of Loss of Memory

6
 STAT 2001 Tutorial 3

Exercise 7. Suppose Red Rose tea randomly places 1 of 3 English porcelain miniature animals
in a box of the tea. On the average, how many boxes of tea must be purchased by a customer
to obtain a complete collection consisting of 3 different animals?

5 Negative Binomial Distribution

Summary
Let X ∼ N B(r, p).
!
x−1 r
1. The p.m.f: f (x) = p (1 − p)x−r ,x = 1, 2, ..., n.
r−1
r r(1 − p)
2. E(X) = , V ar(X) =
p p2
Remark 3. When r = 1, N B(1, p)= Geom(p)