Decision making under

uncertainty through decision

trees in business
 Table of Contents

Introduction to decision Key concepts and Steps involved in

trees and their role in components of constructing a decision
decision making under decision trees in tree for business
uncertainty business context decision making

Evaluation and analysis Comparison of Case studies and real-

of decision tree decision tree approach world applications of
solutions in business with other methods for decision trees in
scenarios decision making under business
uncertainty
01 Introduction to decision
trees and their role in
decision making under
uncertainty
Understanding decision trees for uncertain
scenarios
Decision trees for visualizing complex decisions
Decision trees are graphical tools that help visualize complex decisions and the
potential outcomes under uncertain scenarios.

Probabilistic modeling with decision trees

Decision trees enable probabilistic modeling to assess the likelihood of different
outcomes, aiding in decision making under uncertainty.

Utilizing decision trees for risk assessment

Decision trees are utilized for risk assessment, providing a structured approach to
evaluate uncertain situations and potential consequences.
Application of decision trees in business strategies

Strategic decision-making using decision trees

Decision trees support strategic decision-making in business by analyzing various courses of action and
their associated risks and rewards.

Resource allocation based on decision tree analysis

Businesses use decision tree analysis to allocate resources effectively by considering the uncertainties
and potential outcomes of different choices.

Forecasting and planning with decision trees

Decision trees aid in forecasting and planning within business strategies by considering various
potential scenarios and their probabilities.
Four Models
Decision Making Under Certainty
Decision Making Under Uncertainty
Decision Making Under Risk
Marginal Analysis with Discrete Distribution
Marginal Analysis with Normal Distribution

Decision making Under Uncertainty

a. The pay off for each alternative
b. There are few state of natures
c. Probability of occurrence of each state of nature is unknown.
Application
Mary is the market analyst of Neste and she was asked to present an expansion
plan for the company to the senior manager, which is either to construct a plant,
open a distribution center or do nothing. She was able to determine the expected
pay off depending on the decision and the type of market, which may be favorable,
Unfavorable, average to the business. If the decision is to construct a plant, See pay
off table.

Mary noted that the decision of each manager is affected by his her personality.
She described each manager as follows:
1. Edward is an optimistic person
2. Agnes is very conservative
3. William thinks there is 75% chance that the market is favorable
4. Jane thinks that each type of market has an equal chance of occurring
5. Joel does not want to regret making a wrong decision.
What would be the decision of each manager? How much gain or loss for the company
Would each expect? Which alternative is most likely to be chosen by most managers?
There are four criteria that may be used.
1. Maximax
2. Maximin
3. Hurwicz of Realism
4. La Place (Equally Likely)
5. Minimax Regret
Pay off Summary
Expansion Plan Pay off Types of Market
(000) Favorable Average Unfavorable
Construct a Plant 8 -1 -5
Open a Distribution Center 4 2 -3
Do nothing 0 0 0
Criterion of Realism 0.75
Set Up the Maximax Criterion

Manager Edward Maximax

Expansion Type of Market
Plan (000)
Favorable Average Unfavorable
Construct 8 -1 -5 8
Open 4 2 -3 4
Distribution
Do Nothing 0 0 0 0
Maximum 8
Decision Construct a Plant
Set Up the Maximin Criterion

Manager Agnes Decision Criterion Maximin

Expansion Type of Market
Plan (000)
Favorable Average Unfavorable
Construct 8 -1 -5 -5
Open 4 2 -3 -3
Distribution
Do Nothing 0 0 0 0
Maximin 0
Decision Do nothing
Set Up the Hurwicz Criterion. This is based on choosing an alterative who
have subjective assessment on the chance that market is favorable
Manager Agnes Decision Criterion Hurwicz
0.75
Expansion Type of Market
Plan (000)
Favorable Average Unfavorable
Construct 8 -1 -5 8x
Open 4 2 -3 -3
Distribution
Do Nothing 0 0 0 0
Maximin 0
Decision Do nothing
Determine the expected pay off per row corresponding each alternative where:
Pf = if the market is favorable
Pu = if the market is unfavorable
Hurwicz constant = 0.75 probability that the market is favorable
1- constant = probability that the market is unfavorable
Hurwicz = constant(pf) + (1- constant)Pu
Manager Wiliam Decision Criterion Hurwicz
0.75
Expansion Plan Type of Market
(000)
Favorab Average Unfavorable
le
Construct 8 -1 -5 4.75
Open Distribution 4 2 -3 2.25
Do Nothing 0 0 0 0
Hurwicz Max 4.75
Decision Construct Plant
 La Place Criterion. Choosing alternative for decision with equal chances
of occurrence
Manager Jane Decision Criterion La Place
Expansion Plan Type of Market
(000)
Favorab Average Unfavorable Average
le
Construct 8 -1 -5 0.67
Open Distribution 4 2 -3 1.00
Do Nothing 0 0 0 0
La Place Max 1
Decision Open a Distribution Center
Minimax criterion. Choosing alternative with minimum losses
Determine the maximum pay off per column corresponding to type of market
Mf = maximum pay off if the market is favorable
Ma = maximum pay off if the market is average
Mu = maximum pay off if the market is unfavorable

Mf = (8,4,0) = 8
Ma = (-1,2,0) = 2
Mu = (-5,-3, 0) = 0

Determine the opportunity loss where:

Of =opportunity loss if market is favorable
Oa = opportunity loss if market is average
Ou = opportunity loss if market is unfavorable
La Place Criterion. Choosing alternative for decision with equal chances
of occurrence
Manager Jane Decision Criterion Minimax
Expansion Plan Type of Market
(000)
Favorab Average Unfavorable
le
Construct 8 -1 -5
Open Distribution 4 2 -3
Do Nothing 0 0 0
Maximum 8 2 0
Opportunity Loss
Manager Jane Decision Criterion Minimum Loss
Opportunity Loss
Expansion Plan (000) Type of Market
Favorable Average Unfavorable
Construct 0 3 5 5
Open Distribution 4 0 3 4
Do Nothing 8 2 0 8
Minimum 4
Decision Open
Distributor
Decision Making Under Risk

Conditions:
1. The pay off for each alternative is known
2. There are few state of natures
3. Probability of occurrence of each state of nature is known
4. The cost of perfect information is known
Manager Jane Decision Criterion
Opportunity Loss
Expansion Plan (000) Type of Market
Favorable Average Unfavorable
Construct 0 3 5
Open Distribution 4 0 3
Do Nothing 8 2 0
Probability 0.5 0.2 0.3
Cost of Perfect 1.50
Information
Manager Jane Decision Criterion
Opportunity Loss
Expansion Plan (000) Type of Market
Favorable Average Unfavorable EMV
Construct 0 3 5 2.3
Open Distribution 4 0 3 1.5
Do Nothing 8 2 0 0
Probability 0.5 0.2 0.3
Max 2.3
Decision Contsruct a Plant
Manager Decision Criterion
Opportunity Loss
Expansion Plan (000) Type of Market
Favorabl Average Unfavorable EMV
e
Construct 0 3 5 2.3
Open Distribution 4 0 3 1.5
Do Nothing 8 2 0 0
Max 2.30
Decision Construct a Plant
Probability 0.5 0.2 0.3
EVwPI 4 .4 0 4.4
Cost of Perfect Information 1.5 2.10
Net Gain with Perfect Information 0.6
 Benefits and limitations of decision
trees in decision making

Advantages of decision Challenges in decision Integration of decision

trees for decision making making using decision trees with other decision-
trees making tools
Decision trees offer transparency, Decision trees may oversimplify Exploring the integration of decision
simplicity, and the ability to handle complex scenarios and require trees with other decision-making
complex decision scenarios, making accurate input data, posing tools to enhance the accuracy and
them valuable for decision making challenges in certain decision- reliability of decision outcomes.
under uncertainty. making contexts.
 02 Key concepts and
components of decision
trees in business context
Understanding the decision tree structure

Nodes, branches, and leaves

Decision trees consist of nodes, which represent decision points, branches, which
depict possible outcomes, and leaves, which signify the final decision or outcome.

Splitting criteria and information gain

The process of determining the best split at each node based on criteria such as
entropy and information gain, which helps in maximizing the homogeneity of the
nodes.

Decision rules and paths

The rules that guide the traversal of the tree, leading to specific paths and ultimately,
the decision or prediction.
 Utilizing decision trees for business decisions

Predictive analysis and forecasting

Using historical data and variables to predict future outcomes, aiding in strategic planning and forecasting.

Risk assessment and management

Evaluating potential risks associated with business decisions and identifying risk-mitigation strategies
through scenario analysis.

Resource allocation optimization

Determining the optimal allocation of resources based on various decision paths, maximizing
efficiency and cost-effectiveness.
 Challenges and considerations in
decision tree implementation

Overfitting and model Data quality and feature Interpreting and

complexity selection communicating results
The risk of creating a model that is Ensuring the quality of input data and Effectively communicating the
overly complex and tailored to the selecting relevant features to build insights derived from decision trees
training data, leading to poor an accurate and robust decision tree to stakeholders and decision-
generalization and inaccurate model. makers, ensuring proper
predictions. understanding and action.
 03 Steps involved in
constructing a decision tree
for business decision making
Data Collection and Analysis

Gather relevant data

Collect data from various sources to ensure comprehensive analysis.

Perform data analysis

Analyze the collected data to identify patterns and trends.

Identify relevant variables

Determine the key variables that impact the decision-making process.
 Decision Tree Construction

Select the appropriate algorithm

Choose the suitable decision tree algorithm based on the nature of the problem.

Construct the initial tree

Build the initial decision tree based on the selected algorithm and variables.

Prune the tree (if necessary)

Refine the decision tree by eliminating unnecessary branches to improve accuracy.
 Evaluation and Implementation

Evaluate the decision tree Implement the decision Monitor and update the
tree tree
Assess the effectiveness of the Incorporate the decision tree into the Regularly review and update the
constructed decision tree through decision-making process to guide decision tree to ensure its relevance
testing and validation. strategic choices. and accuracy.
04 Evaluation and analysis of
decision tree solutions in
business scenarios
 Understanding the concept of decision trees

Decision trees in Benefits of decision trees Limitations of decision Real-world applications

business trees

Decision trees are a visual Exploring the advantages of Analyzing the constraints and Examining real-life instances
representation of potential decision trees in business potential drawbacks of where decision trees have
decision paths and outcomes scenarios, including their employing decision trees in been effectively utilized to
in a business context, aiding in ability to handle uncertainty complex and dynamic address intricate business
strategic planning and risk and facilitate informed business environments. challenges and opportunities.
management. decision-making.
 Implementing decision tree solutions

Data preparation and analysis Model development and

validation

Exploring the crucial steps involved in preparing and The process of building and validating decision tree
analyzing data to construct decision trees for models to ensure their accuracy and reliability in
business decision support. diverse business contexts.

Integration with business Performance monitoring and

strategies refinement

Strategies for integrating decision tree solutions The importance of continuously monitoring decision
seamlessly with existing business strategies and tree performance and refining the models to align with
operational processes for maximum impact. evolving business requirements.
 Interpreting decision tree results
Decision tree
Strategies for interpreting and communicating decision tree results to stakeholders
interpretability and and decision-makers, ensuring comprehension and buy-in.
communication
Visual aids and storytelling
Utilizing visual aids and storytelling techniques to effectively convey decision tree
insights and recommendations to diverse business audiences.

Addressing uncertainty and sensitivity

Managing uncertainty and sensitivity within decision tree interpretations and
communications to foster confidence and trust in the outcomes.

Adapting to audience preferences

Tailoring decision tree communication approaches to align with the preferences and
cognitive styles of different business stakeholders.
 Challenges and considerations in decision tree
implementation

Data quality and reliability Complexity and scalability Ethical and legal Human bias and cognitive
implications factors

Addressing the challenges Navigating the complexities Considering the ethical and Examining the influence of
related to data quality and and scalability concerns legal implications of decision human bias and cognitive
reliability when utilizing associated with deploying tree-based decision-making, factors on decision tree
decision trees for critical decision tree solutions across particularly in sensitive outcomes and ensuring
business decisions and diverse business operations business domains and social mitigation strategies are in
strategies. and scenarios. contexts. place.
 Future advancements and opportunities

Innovations in decision tree Integration with AI and predictive

technology analytics

Exploring emerging technological advancements The potential synergies and opportunities presented
and innovations shaping the future of decision trees by integrating decision trees with artificial intelligence
in revolutionizing business decision-making. and predictive analytics for enhanced business
insights.

Expanding application Training and skill development

domains

Assessing the expanding application domains of The importance of investing in training and skill
decision trees beyond traditional business contexts, development to harness the full potential of decision
including healthcare, finance, and environmental tree technologies and tools in diverse business
sustainability. landscapes.
 05 Comparison of decision
tree approach with other
methods for decision making
under uncertainty
Accuracy and Simplicity

Decision Tree Approach

Decision trees offer a balance between accuracy and simplicity, making

them easy to interpret and implement.

Other Methods

Other methods may sacrifice simplicity for accuracy or vice versa, leading to
challenges in practical application.
Handling Non-linear Relationships

Decision Tree Approach

Decision trees can effectively handle non-linear relationships in the data,

offering flexibility in modeling.

Other Methods

Other methods may struggle with capturing non-linear relationships, limiting

their modeling capabilities.
Interpretability and Transparency

Decision Tree Approach

Decision trees provide transparent decision-making processes, allowing

stakeholders to understand the rationale behind choices.

Other Methods

Other methods may lack interpretability, leading to challenges in explaining

decisions to stakeholders.
Data Preprocessing Requirements

Decision Tree Approach

Decision trees can handle both numerical and categorical data without
extensive preprocessing, saving time and effort.

Other Methods

Other methods may require extensive data preprocessing, adding

complexity and potential errors to the process.
Handling Missing Values

Decision Tree Approach

Decision trees can handle missing values in the dataset without the need for
imputation techniques, simplifying data preparation.

Other Methods

Other methods may require complex imputation techniques for handling

missing values, adding extra complexity to the process.
06 Case studies and real-
world applications of
decision trees in business
 Market trend analysis for product launch
Identifying market
Decision trees help in identifying patterns and predicting customer preferences,
trends and customer aiding in successful product launches.
behavior
Customer segmentation for targeted marketing
Segmenting customers based on behavior and preferences to tailor marketing
strategies for better engagement and conversion.

Risk assessment for investment decisions

Using decision trees to assess potential risks and returns, aiding in informed
investment decisions for business growth.

Optimizing supply chain and inventory management

Utilizing decision trees to forecast demand and optimize inventory levels, leading to
cost savings and improved efficiency.
 Enhancing operational efficiency and resource allocation

Workforce planning and Resource allocation for Predictive maintenance Optimal pricing strategies
skill assessment project management for equipment and assets and revenue optimization

Decision trees aid in Optimizing resource allocation Using decision trees to predict Analyzing customer data to
identifying skill gaps and based on project requirements maintenance needs, reducing determine optimal pricing
planning workforce allocation and constraints, leading to downtime and extending asset strategies for maximizing
for optimal operational efficient project delivery. lifespan for cost savings. revenue and profitability.
efficiency.
 Improving customer service and satisfaction

Personalized recommendations Churn prediction and proactive

and product bundling retention strategies

Creating personalized product recommendations Identifying potential churn indicators to proactively

and bundling based on customer preferences and implement retention strategies and enhance customer
purchase history. loyalty.

Complaint resolution and Customer lifetime value

service escalation prediction

Using decision trees to streamline complaint Predicting the potential value of customers to tailor
resolution processes and efficiently escalate retention and engagement strategies for maximizing
service requests. lifetime value.
 Fraud detection and risk mitigation
Mitigating operational
Identifying fraudulent activities and mitigating risks through data-driven decision
risks and optimizing trees for enhanced security.
decision-making
Operational bottleneck identification and resolution
Using decision trees to identify operational bottlenecks and implement targeted
solutions for improved efficiency.

Regulatory compliance and decision support

Ensuring compliance with regulations through decision trees to support informed
and compliant decision-making.

Crisis management and scenario planning

Utilizing decision trees to plan for potential crises and develop effective strategies for
crisis management and resilience.
 Enabling data-driven strategic decision-making

Market expansion and Competitive intelligence Strategic partnership Mergers and acquisitions
diversification analysis and market positioning identification and decision support
evaluation
Using decision trees to Gaining insights into Identifying potential strategic Utilizing decision trees to
analyze market opportunities competitors' strategies and partners and evaluating assess potential mergers and
and potential diversification market positioning through collaboration opportunities acquisitions, aiding in strategic
strategies for business growth. decision tree analysis for through decision tree analysis. business expansion.
informed decision-making.
Thank You
Smallppt