TOPIC 4: DATA ANALYSIS

Focus Topics:
●​ Decision-making under certainty, risk, and uncertainty
●​ Decision trees and payoff tables

Decision Analysis
-​ is a structured process for making better decisions by evaluating different options, their
outcomes, and the factors involved — especially when dealing with certainty, risk, or uncertainty.
Decision making
-​ is the process of choosing the best option from available choices to solve a problem or reach a
goal.
-​ Every decision involves a level of information and outcome expectation. This leads us to three
main aspects of decision-making: certainty, risk, and uncertainty.

Let’s compare!

Aspect Under Under Risk Under Uncertainty

Certainty

Outcomes Known Known Known

Probabilities Predictable. Known or can be Unknown or cannot be

Only one possible outcome. reliably estimated reliably estimated
No doubt.

Information All information is fully Partial information Very limited or no reliable

known.Accurate, measurable, available information
and reliable.

Decision Basis Complete information. Facts, Data, models, and Assumptions, judgment, or
outcomes are known with 100% probability analysis decision criteria
accuracy. No need for
probabilities and assumptions.

Common Tools Simple Comparisons, Expected value, Maximin, Maximax, and

Cost-benefit analysis, decision trees, Minimax Regret
Break-even Analysis, Linear Return-Risk Trade Off
Programming. (simulation)
Decision Making under – CERTAINTY 2.​ COST-BENEFIT ANALYSIS
Definition: a systematic process of
-​ The decision-maker knows exactly comparing the total expected costs of a
what will happen for each course of decision or project with the total expected
action. benefits, to determine whether it is
-​ Under conditions of certainty, financially worthwhile.
accurate, measurable, and reliable
information and knowledge on Example Scenario: A restaurant considers
which you base your decisions are buying a fryer.
available to you. Cost: P100,000
-​ The future and outcome are highly Annual savings (oil, labor): ₽50,000
predictable under conditions of Useful life: 3 years
certainty.
Total Benefit = ₽50,000 × 3 = ₽150,000
Net Benefit = P150,000 - ₽100,000
●​ Examples using the common tools =₽50,000
under Certainty(definition) :
• Decision: Buy the fryer, as the financial
1.​ SIMPLE COMPARISONS gain exceeds the cost.
Definition: Compare known values (cost,
time, speed) of options directly.

Example Scenario: A retail store must 3.​ Break- Even Analysis

choose between two cash register systems. Definition: It shows the point where total
revenues equal total costs—this is called
• Option A: ₱40,000, lasts 5 years the break-even point. At this point, the
• Option B: ₱45,000, lasts 6 years business is not losing money but also
not making a profit yet.
Option A: ₱40,000 ÷ 5 = ₱8,000/year
Option B: ₱45,000 ÷ 6 = ₱7,500/year Formula:
Break-Even Point (units)=
Fixed cost
So, Option B gives more value per year,
Selling Price per Unit - Variable Cost per Unit
₱500 less each year than Option A

• Decision: Example Scenario: A coffee shop launches

-Based on total value and lifespan, Option a new drink.
B offers longer use with minimal added Fixed Costs: P80,000
cost. Selling Price: P120
-Choose Option B if budget allows and you Variable Cost: P60
value durability.
Break-even = ₽80,000 / (P120 - P60)
= 1,333 cups
Decision: Sell at least 1,333 cups to start (-4x + 2y) - (x + 2y) = 200 - 100
(cancel out both 2y)
making a profit.
3х = 100
(Divide both side by 3)

4.​ LINEAR PROGRAMMING Solve for x:

Definition: It helps you maximize x= 100 / 3 = 33.33
something (like profit) or minimize Intersection point: (x, y) = (33.33, 33.33)
something (like cost) - while staying within Meaning:
your limits (like budget, time, or materials). Make 33.33 breads and 33.33 muffins to
fully maximize use of labor and flour.
Example Scenario: a bakery produces
bread and muffins to maximize profit.
Bread: P30 profit, requires 2 hrs labor & 1
kg flour Decision Making under – RISK
Muffin: P20 profit, requires 1 hr labor & 2 kg
flour -​ The process of choosing between
Available: 100 hrs labor, 100 kg flour alternatives when the possible
outcomes are known, and each
Let: Labor Flour outcome has a known or predicable
x= Bread 2hrs 1kg probability of occurring.
y= Muffins 1 hr 2kg -​ It involves evaluating the potential
consequences of each option and
Constraints Equation: selecting the one that offers the
●​ 2x + y ≤ 100 (Labor) most favorable expected result
●​ x + 2y ≤ 100 (Flour) based on these probabilities.

To find the intersection point, we convert: ➔​ Key Characteristics:

- 2x + y = 100 ●​ Known possible outcomes
- x + 2y = 100 ●​ Probabilities assigned to
(We use the equal sign (=)(instead of the “≤” sign) because
we're solving for the exact point where the two limits each outcome
(constraints) are used up completely. Instead of using the ●​ Quantifiable and measurable
diagram)
risk
We will use elimination to solve these ●​ Goal - Oriented
equations.

Make Coefficients Match

Multiply equation (1st): (2)2x + y = 100(2) ➔​ Examples
(mul:by 2) 1.​ Launching a New Product
Result: 4x + 2y = 200 ❖​ A company plans to launch a
Now we have: new gadget.
- 4x + 2y = 200 ❖​ Probabilities:
- x + 2y = 100 ➢​ 50% big hit – earn $5
million
Subtract both of the Equations:
 ➢​ 30% perform
moderately – $2 million
➢​ 20% will fail – $1 million
loss
❖​ Based on these probabilities Where:

and outcomes, the company ●​ Pi = Probability of outcome iii

decides whether to launch ●​ Oi​= Outcome value iii
the product or not. ●​ n = Number of possible outcomes

Or:

EV = (Outcome 1 × Probability 1) +
2.​ Expanding to a New Market
(Outcome 2 × Probability 2) + …
❖​ A restaurant chain considers
opening in a new city.
❖​ Market research shows: 2.​ Decision Tree Analysis
➢​ 60% chance the expansion -​ This is a visual tool that helps you
will increase profits by $3 map out every possible decision,
million outcome, and probability.
➢​ 25% chance of breaking
even 3.​ Risk-Return Trade-Off
➢​ 15% chance of losing $1 -​ An approach of balancing risk with
million due to low customer reward
demand. -​ Choose based on your “risk
❖​ The company uses these appetite” – are you cautious or bold?
probabilities to decide if the -​ higher potential returns come with
expansion is worth the risk. higher risk, while lower risk usually
means lower returns.
“This is decision making under risk because
the business is using data, probabilities, and
clear outcomes to make a smart, rational,
goal-based choice.”
Decision Making under – UNCERTAINTY

-​ The process of choosing between

➔​ Approaches different options when the possible
1.​ Expected Value (EV) Approach outcomes are known, but there is
-​ The most common method no reliable way to determine how
-​ It involves multiplying each possible likely each outcome is.
outcome by its probability, then -​ Since the probabilities are
adding the results. unknown, decisions are usually
-​ The option with the highest based on judgment, experience,
expected value is usually or general decision rules rather
considered the best choice. than data.
-​ Formula:
➔​ KEY CHARACTERISTICS "Deciding with a map that has no
●​ Known possible outcomes roads—just destinations."
●​ Unknown probabilities
●​ Lack of Information
●​ Reliance on Judgment or
Assumptions ➔​ Approaches
●​ Higher Level of Risk 1.​ Maximin (Pessimistic Approach)
-​ Choose the option with the best
worst-case outcome.
-​ Assumes the decision-maker wants
➔​ EXAMPLES to avoid the worst.
1.​ Starting a New Business in an -​ Example: “Even if things go badly, I
Unfamiliar Industry want the safest result possible.”
❖​ Someone opens a new café in a 2.​ Maximax (Optimistic Approach)
town where they’ve never lived -​ Choose the option with the best
before. possible payoff.
❖​ Outcomes: -​ Assume everything will go as well
➢​ succeed as it can.
➢​ break even -​ Example: “I’ll go for the highest
➢​ fail reward, even if it’s risky.”
❖​ But they don’t know how many
customers will actually come, so
they can’t predict the outcome.
❖​ They have to make a decision 3.​ Minimax (Regret Theory)
based on gut feeling and hope, -​ Choose the option that minimizes
not solid numbers. the maximum possible regret.
-​ Focuses on avoiding the feeling of “I
should have chosen the other one.”
-​ Example: “Which choice will I regret
2.​ Introducing/Selling a Product in a the least if it goes badly?”
New Country
❖​ A company launches a product in a
country where it has never operated
before. Decision Tree
❖​ Outcomes
➢​ Success -​ A support tool with a tree-like
➢​ Slow growth structure that models probable
➢​ Failure outcomes, cost of resources,
❖​ But they don’t really know the utilities, and possible consequences.
tastes or habits of the people Decision trees provide a way to
there, so there’s no clear data to present algorithms with conditional
rely on. control statements. They include
❖​ They have to guess and take a branches that represent
chance. decision-making steps that can lead
to a favorable result.
-​ There a two primary types of showing either an action you
Decision trees: The Classification take or an event's outcome.
tree and The Regression trees

-​ Classification trees classifies data

in distinct categories or classes
meanwhile Regression Trees are
used when data are in numerical How to Analyze a Decision Tree (Working
value Backwards):

-​ The main way to figure out the best

path is by working backward from
the end of the tree to the start.

➔​ KEY PARTS OF A DECISION ~ At Chance Nodes: Calculate the Expected

TREE: Monetary Value (EMV). This is done by
◆​ Decision Nodes (Squares): A multiplying each outcome's payoff by its
decision node is a node in an probability (as a decimal), and then adding
activity at which the flow all those results together.
branches into several
optional flows. There is EMV=∑(Pi×Oi)
exactly one incoming edge
Where Pi is the probability of an outcome,
and an arbitrary number of
and Oi is the result of an outcome
outgoing edges, which each
have a condition. ~ At Decision Nodes, Choose the option
◆​ Chance Nodes (Circles): that gives you the best EMV (highest for
These are points where an profits, lowest for costs). The EMV of this
uncertain event happens. chosen option then becomes the value for
The lines from these circles that decision node.
show the possible results of
that event, and each has a Repeat this process until you reach the very
percentage probability (how first decision, which will show you the best
likely it is to happen). All overall strategy.
probabilities from one circle
must add up to 100%. Example: Product Launch Decision Tree
◆​ Outcome Nodes (Payoffs): Analysis
These are the end results at
the end of any path through Scenario: A company is deciding whether to
the tree. They show the launch a new product. They can launch it
profit, cost, or other value. now or pay for market research first.
◆​ Branches: These are the Research costs ₱10,000 and can be
lines that connect everything, Favorable (60% chance) or Unfavorable
(40% chance). The product's success (High
or Low Demand) depends on these choices.
Unfavorable Research (P=40%)

Decision 2.2: (Launch or Don't Launch)

Option 2.2.1: Launch

Chance Node: Market Demand

High Demand (P=30%): Payoff = ₱100,000

Low Demand (P=70%): Payoff = ₱20,000

Option 2.2.2: Don't Launch: Payoff = ₱0

Analysis (Working Backwards):

Decision 1: (Launch Now or Do Research)

Option 1: Launch Now

Analyze Chance Node (Launch Now):
Chance Node: Market Demand

High Demand (P=70%): Payoff = ₱200,000

EMV (Launch Now) = (0.70 × ₱200,000) +
(0.30 × ₱50,000)
Low Demand (P=30%): Payoff = ₱50,000
EMV (Launch Now) = ₱140,000 + ₱15,000
= ₱155,000
Option 2: Do Market Research (Cost =
Analyze Chance Node (Launch after
₱10,000)
Favorable Research):
Chance Node: Research Outcome

Favorable Research (P=60%)

EMV (Launch | Favorable) = (0.80 ×
Decision 2.1: (Launch or Don't Launch) ₱250,000) + (0.20 × ₱60,000)

Option 2.1.1: Launch EMV (Launch | Favorable) = ₱200,000 +

₱12,000 = ₱212,000
Chance Node: Market Demand
Analyze Decision Node 2.1 (After
High Demand (P=80%): Payoff = ₱250,000 Favorable Research):

Low Demand (P=20%): Payoff = ₱60,000

Option 2.1.2: Don't Launch: Payoff = ₱0 Compare: Launch (₱212,000) vs. Don't
Launch (₱0)
Decision: Choose Launch. Value at Node Compare: Launch Now (₱155,000) vs. Do
= ₱212,000 Market Research (₱134,800)

Analyze Chance Node (Launch after Decision: Choose Launch Now.

Unfavorable Research):
Optimal Strategy: The best decision is to
Launch Immediately, as it results in the
highest expected value of ₱155,000.
EMV (Launch | Unfavorable) = (0.30 ×
₱100,000) + (0.70 × ₱20,000) A Payoff table, also called a Payoff matrix,
is a table used in decision analysis to show
EMV (Launch | Unfavorable) = ₱30,000 + outcomes (payoff) for different decisions
₱14,000 = ₱44,000 under various states of nature (possible
future events). It helps decision makers
Analyze Decision Node 2.2 (After
compare options based on different
Unfavorable Research):
scenarios.

Parts of a Payoff Table:

Comparing them between: Launch ●​ Decision Alternatives (Rows): The
(₱44,000) vs. Don't Launch (₱0) actions you can choose from.
●​ States of Nature (columns): The
Decision: Choose Launch. Value at Node possible situations or outcomes in
= ₱44,000 the future.
●​ Payoffs (Cells): The result (e.i.,
Analyze Chance Node (Research profit, cost, utility) of choosing a
Outcome): particular action in a particular state.

Example of a Simple Payoff Table:

EMV (Market Research Outcome) = (0.60 ×
Value from D2.1) + (0.40 × Value from D2.2) Decision State of State of State of
Alternati Nature A Nature B Nature
EMV (Market Research Outcome) = (0.60 × ves C
₱212,000) + (0.40 × ₱44,000)
Option 1 100 50 -30
EMV (Market Research Outcome) = Option 2 70 60 40
₱127,200 + ₱17,600 = ₱144,800
Option 3 30 80 50
Now, subtract the initial research cost:
₱144,800 - ₱10,000 = ₱134,800
Decision making methods used in payoff
tables
Analyze Decision Node 1 (Initial
1.​ Maximax (Optimistic)- “Go for the
Decision):
Biggest win”. Choose the option with
the highest possible gain.
 2.​ Maximin (Conservative or
Mutual Funds 17 0 10
Pessimistic)- Choose the option with
the best worst-case scenario. Maximum= Bonds= 30, Stocks= 18, Mutual
3.​ Minimax Regret- Choose the best Funds= 17
option that has the least regret if you Minimum= 17 (Mutual Funds)
make the wrong choice.
4.​ Expected Value- Choose the option Expected (Monetary) Value
with the highest average result, Alternatives Growing Stable Declining
based on the probabilities of what
might happen. Bonds 40 45 5

Stocks 70 30 -13
DECISION TABLE
Alternatives Growi Stable Declini Mutual 53 45 -5
ng ng Funds

Bonds 40 45 5 Probability 0.2 0.50 0.3

Stocks 70 30 -13 EMV(Bonds) = 0.2(40) + 0.5(45) + 0.3 (5) =

32.0
Mutual Funds 53 45 -5
EMV(Stocks) = 0.2(70) + 0.5(30) + 0.3 (-13)
= 25.1
Maximax (Optimistic) Approach = Invest EMV(Mutual Funds)= 0.2(53) + 0.5(45) +
in Stocks 0.3(-5) = 31. 6
(Best of Best) Bonds= 45, Stocks= 70, Decision: Invest in Bonds
Mutual Funds= 53
Expected Value of Perfect Information-
Maximin Approach= Invest in Bonds maximum payment for additional
(Best of Worst) Bonds= 5, Stocks= -13, information
Mutual Funds= -5 EVPI = EVwPI - EVwoPI
EVwoPI = 32 (Maximum EMV, see the
Minimax Regret= Invest in Mutual Funds solution above)
(Minimizes the Maximum Regret) Regret= EVwPI = 70(0.2) + 45(0.5) + 5(0.3) = 38
Best Payoff - Payoff Received EVPI= 38 - 32
Bonds= 70 - 40 = 30, 45 - 45 = 0, 5 - 5 = 0 EVPI= 6
Stocks= 70 - 70 = 0, 45 - 30 = 15, 5 - (-13)
= 18
Mutual Funds= 70 - 53 = 17, 45 - 45 = 0, 5-
(-5) = 10
REGRET TABLE
Alternatives Growi Stable Declin
ng ing

Bonds 30 0 0

Stocks 0 15 18