DOTE 2011 | Fall 2024

@ CUHK Business School

Statistical Analysis for Business Decisions

Probability

Yunduan Lin
Assistant Professor
Department of Decisions, Operations and Technology
CUHK Business School
 Agenda

Statistical Analysis for Business Decisions

01 Classical Definition of Probability

o Operations on events
o Special events

02 Axiomatic Definition of Probability

o Basic rules
o Conditional probability
o Independence
 Probability - Definition
Probability
Long-run relative frequency
A possibility of random experiment
Repeat a random experiment many many times
A group of outcomes

o Something is going to occur

o Know what may happen but no idea which one
Toss a fair coin one million times: o It may repeat itself for many times
o Roughly half a million outcomes are head
o Roughly half a million outcomes are tails
o Probability of head is ½, for tail is also ½
 Probability - Example
Rolling a fair dice

o Random experiments: repeatedly rolling this dice.

o Outcomes: 1 is an outcome; 2,3,4,5,6 are also outcomes.
o Sample space: is the set of all outcomes.
o Event:
• Even numbers
• Odd numbers
• Large numbers
• You can define others …

o Probability:
 Probability - Example
Rolling a fair dice twice

o Random experiments: repeatedly rolling a dice twice.

o Outcomes: an outcome can be represented by a pair of integers
• The first entry is the number of the first trial, second entry is the number of the second trial
• (1,2) and (2,1) are different outcomes
• Total number of outcomes is 36.
o Sample space:
o Event:
• The sum of two trials is less than 4
• The numbers of two trials are the same

o Probability:
 Operation on Events - Intersection
Intersection

An event that contains all common outcomes in event A and event B

Example: Rolling a fair dice

o Event:
• Even numbers
• Odd numbers
• Large numbers
 Operation on Events - Union
Union

An event that contains all common outcomes in event A and event B

Example: Rolling a fair dice

o Event:
• Even numbers
• Odd numbers
• Large numbers
 Operation on Events - Complement
Complement

An event that contains all outcomes not in event A (but in sample space)

Example: Rolling a fair dice

o Event:
• Even numbers
• Odd numbers
• Large numbers
 Special Events – Mutually Exclusive Events
Mutually Exclusive Events
Events that cannot happen at the same time: intersection is empty.

Events A and B are mutually exclusive if

Example: Rolling a fair dice

o Event:
Mutually exclusive events
• Even numbers
• Odd numbers
• Large numbers

At most one of the events will happen.

 Special Events – Exhaustive Events
Exhaustive Events
One of the events must happen: union of all events = the sample space.

Events A and B are exhaustive if

Events A1, A2, …, An are (collectively) exhaustive if

Example: Rolling a fair dice

o Event:
Exhaustive events
• Even numbers
• Odd numbers
• Large numbers

At least one of the events will happen.

 Special Events – Partition
Partition
Partition of sample space: Mutually exclusive + Exhaustive events.

Events A1, A2, …, An is a partition of sample space if they are mutually exclusive
and exhaustive

Example: Rolling a fair dice

o Event:
A partition of the sample space S
• Even numbers
• Odd numbers
• Large numbers

One and only one of the events will happen.

 Axiomatic Definition of Probability
Probability
A function from events to numbers that satisfies the following three axioms.

o Axiom 1: For any event A, we have Always between 0 and 1 since it is a fraction

o Axiom 2: For sample space S, we have

o Axiom 3: If event A is a collection of n mutually exclusive outcomes

, then we have

For rolling a fair dice, the probability of even numbers is the sum of probabilities of having 2, 4 and 6.
 Complement Rule - Definition
Complement Rule
Useful for a complicated event but
Let be the complement of A. Then simple complement

Also implies

Proof

First, are mutually exclusive and exhaustive events.

By Axiom 3, we have

By Axiom 2, we have

Hence, we have
 Complement Rule - Example
Complement Rule Example: Rolling a Dice

What is the probability that the face value is less than 6?

Let A be the event that it is less than 6.

The complement is the event that it is no less than 6, which has only one outcome. Hence,

Applying the complement rule, we have

 Complement Rule - Extension
Complement Rule - Extension

The same concept can be applied to partition of the sample space.

If is a partition of sample space S, then

Example: Consider rolling a fair dice twice. What is the probability of having the sum of two results
is less than 11?
o Let Ai be the sum of the two results is i.
o Then is a partition of the sample space.
o Hence,
 Addition Rule - Definition
Axiom 3 says if A and B are mutually exclusive, then

What happen when they are not mutually exclusive, namely,

Addition Rule

The probability of A or B is the sum of probabilities A and B net of the probability of A and B:

Addition rule "connects" probabilities of union and intersections.

 Addition Rule - Example
Example: Consider rolling a fair dice once. What is the probability of having an even
number or a number greater than 3?

o Let A = even number. That is, A = {2, 4, 6}.

o Let B = greater than 3. That is, B = {4, 5, 6}.

o Then, A or B = even number or greater than 3.

o Alternatively, A and B = even number and greater than 3.

 Total Probability Rule - Definition

Total Probability Rule

For any events A and B, we have

Example: Consider rolling a fair dice once. What is the probability of having an even
number or a number greater than 3?
A with B

A without B
 Total Probability Rule - Extension

General Total Probability Rule

Similarly, we can extend the total probability rule from complements to partition:

If A1, A2, …, An is a partition of sample space S, then

 Conditional Probability - Definition

o Update probability when we know something is true.

o Since the sample space changes, we have to update it to make sure the sum of
probabilities of all outcomes is one.

Conditional Probability

Conditional probability of event A given event B has happened

Known fact

Given / condition on
 Conditional Probability - Examples
Example: Consider rolling a fair dice once. Suppose we already know that the outcome of
dice is large. That is, we know that L = {4, 5, 6} is true. What is the probability that the result
is an even number given this?

o The sample space of the original problem is S = {1, 2, 3, 4, 5, 6}. The event of even
number is E = {2, 4, 6}.
Definition: all outcomes
o When we know that L is true, we can update the sample space to L = {4, 5, 6}.
Accordingly, the event of even number within L should also be updated to {2, 4, 6}.

o Hence,

o Alternatively, we can have

 Multiplication Rule - Definition

Multiplication Rule
Rearrange the definition of conditional probability

The probability of A and B is equal to:

o Probability of A given B times probability of B, or
o Probability of B given A times probability of A
 Multiplication Rule - Examples
Example: Consider an urn with 3 black and 3 white balls. Suppose 2 balls are randomly drawn from
the urn (without replacement). What is the probability that both balls are black?

o Event: B1 = 1st ball is black, B2 = 2nd ball is black

o Pr(B1) = 3/6, Pr(B2|B1) = 2/5

o Hence, we have the probability that both balls are black as

 Multiplication Rule - Examples
Example: Consider an urn with 3 black and 3 white balls. Suppose 2 balls are randomly drawn from
the urn (without replacement). What is the probability that both balls are black?

o Tree diagram
 Multiplication Rule - Examples
Example: An insurance agent wants to study customers' perceptions of its new insurance plan. It
uses a telephone survey in which numbers are called at random. If 30 percent of customers will
actually answer the call, and 10 percent of whom answered are interested in doing the survey. What
is the probability that the agent can gather a survey response when they makes the phone call?

o Event: A = the person answers the phone call, B = the person is interested in doing the survey

o Pr(A) = 30%, Pr(B|A) = 10%

o Hence, we have the probability that the agent can gather a response as
 Multiplication Rule - Extension

Total Probability and Multiplication Rules

Combining the two rules

For any events A and B, we have

If B1, B2, …, Bn is a partition of sample space S, then

 Multiplication Rule - Examples
Example: On an e-commerce platform, 40% of the products are electronics, 30% are clothes, 20%
are food and 10% books. If the proportion of 3-party products for these categories are 30%, 2%,
50%, and 10%, respectively, what is the overall share of 3-party products on this platform?

o Event: B1 = product is electronics, B2 = product is cloth, B3 = product is food, B4 = product is

book, A = product is a 3-party product

o Hence, we have the overall share of 3-party products as

 Multiplication Rule - Examples
Example: One bag (called Bag A) contains 4 white and 1 green balls, another bag (called Bag B)
contains 1 white and 1 green balls. One bag is chosen at random and one ball is drawn at random
from the bag. Find the probability that it is a white ball.

o Event: W = the ball is white, A = the bag is A, B = the bag is B

o Hence, we have the probability that it is a white ball as

 Multiplication Rule - Examples
Example: One bag (called Bag A) contains 4 white and 1 green balls, another bag (called Bag B)
contains 1 white and 1 green balls. One bag is chosen at random and one ball is drawn at random
from the bag. Find the probability that it is a white ball.

o Tree diagram
 Independence - Definition

Statistical Independence
Events are independent if conditional probabilities are the same as those unconditionals.

Events A and B are (statistically) independent if and only if one of the following is true:

o or

o or ​

o .
 Independence - Examples

Sampling With or Without Replacement

Consider drawing two balls from an urn.

o Putting back the first ball, the second draw is independent of the first draw.
o If not putting back, the two draws are dependent.

Sampling with replacement makes each draw independent.

 Independence - Examples

Example: Which are likely to be independent? If two events are correlated with each
other, then they are dependent for sure.
o A. The opening and closing price of a stock.
o B. Taking a taxi home and driver is playing your favorite music.
o C. The fare of MTR and the price the pizza from the Pizza Hut.
o D. The weather condition and the delivery time of HKTV Mall.

Pizza Principle - a humorous but generally historically accurate "economic law"

As reported by The New York Times in 1980, that from the early 1960s "the price of a slice of
pizza has matched, with uncanny precision, the cost of a New York subway ride”.
---Eric M. Bram

MTR is too expensive, let’s stop buying pizza!

 Independence - Examples
Spurious correlation is common. Check a fun website.

AI explanation: The increase in Batman searches led to a rise in vigilante activity across Oklahoma, prompting
a need for more security to handle the caped crusaders creating chaos. Looks like the dark knight really
sparked a guard-ing presence in the state.
 Independence - Examples
Spurious correlation is common. Check a fun website.

AI explanation: The heightened state of unease over the possibility of a 'Con Air' sequel resulted in an influx of
security personnel. It just goes to show, when it comes to Nic Cage, the need for increased security is always
'Raising Arizona.'
 Independence - Examples
Data do lie, data analyst should not

The existence of correlation or dependency does not mean causality!

Clair Cameron Patterson

(1922 – 1995)
Geochemist

Lead Regulation
 Independence - Examples
A Brief History of Lead Regulation
o 1922: Lead was first introduced into gasoline, immediately drawing headlines concerning public health.
The form of lead in gasoline was known as tetraethyl lead and it raised the octane level of gasoline,
resulting in “premium” gas for high-performance engines.

o 1930s: Five workers at a New Jersey plant died, with four of them going “insane” before their death. The
industries rejected scientific evidence, claiming there was no proof of causation and tried to blame the
children and families as being irresponsible for allowing children to eat the paint chips, claiming that they
were “sub-normal to start with.”

o 1965: A geochemist named Clair Patterson in Greenland brought the airborne lead issue into American
consciousness. Until then, industry experts claimed only workers were at risk for lead poisoning, and that
because lead has always been naturally in the air, it must be safe. Using ice core samples, Patterson found
that higher levels of lead existed in recent samples than older ice. He further concluded that the amount of
lead Americans had in their blood was 100 times greater than natural levels.

o 1970: Nixon signed the Clean Air Act of 1970 into law on December 31st. Along with lead, the
Environmental Protection Agency was required to lower emissions of hydrocarbons, carbon monoxide, and
nitrogen oxides by 90 percent in only a few years.
 Independence - Examples
A Brief History of Lead Regulation
 A Feedback Form for the Entire Term

https://docs.google.com/forms/d/e/1FAIpQLSfsEgnMFLypI_KW6GF7j_FXtVY5E4Jrmf2P_BDwaG8GXWDc0A/viewform?usp=sf_link