ACHARYA INSTITUTE OF TECHNOLOGY

Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.

Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

MODULE 4
Chapter 2: Optimization and Simulation
MODEL-BASED DECISION-MAKING
Model-based decision-making refers to making decisions using analytical or mathematical
models.
This is the essence of Prescriptive Analytics, which goes beyond descriptive (what happened)
and predictive (what will happen) analytics to recommend what should be done.
The aim is to optimize outcomes, such as maximizing profit, minimizing costs, or improving
efficiency.

The Role of Prescriptive Analytics

Prescriptive analytics uses analytical models to:
Recommend actions or automate decision-making.
Combine insights from descriptive (past) and predictive (future) analytics to decide the
best course of action.
Examples:
Marketing: Deciding which customers should receive promotional offers to maximize
profit and minimize costs.
Customer Retention: Predicting which customers may churn and offering promotions to
retain them.
Vendor Contracts: Selecting suppliers to minimize cost while ensuring supply needs.
Advertising: Deciding how much to bid for search keywords to maximize ROI.
Operations: Scheduling employees based on customer arrival forecasts.
Logistics: Determining warehouse locations or daily delivery routes to optimize cost and
time.
Model-based decisions are often superior to intuition-based decisions because they are:
More consistent
Data-driven
Justifiable

Importance of Prescriptive Analytics

It represents the next frontier in analytics.
Enables decision analytics — models that guide or automate decisions.
Supported by real-world evidence (e.g., INFORMS publications like Interfaces, ORMS
Today, Analytics Magazine).

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

Example Case:
TurboRouter (Christiansen et al., 2009):
A decision support system for ship routing and scheduling.
Helped a shipping company increase profit by $1–2 million in 3 weeks.

Prescriptive Analytics Model Examples

Modeling is at the heart of prescriptive analytics.
Purpose:
To make data-based recommendations that are computationally manageable for complex
decisions.
Example:
Selecting which customers (from millions) to target with offers to maximize total
response value under a budget constraint → requires a probability-based optimization
model.
Modeling Techniques:
1. Linear Programming – A mathematical technique used to find the optimal solution for
a problem with linear relationships between variables, such as maximizing profit or
minimizing cost.
Example: Determining the best product mix to maximize profit given resource
constraints.
2. Simulation – A method that models real-world processes to observe their behavior
under different scenarios and uncertainties.
Example: Simulating airport operations to reduce passenger waiting time and improve
efficiency.
3. Optimization – The process of selecting the best possible solution from all feasible
alternatives to achieve a specific objective.
Example: Optimizing delivery routes to minimize fuel consumption and travel time.
4. Decision Trees – A graphical tool that helps evaluate alternatives and their possible
outcomes, often used under conditions of risk and uncertainty.
Example: Using a decision tree to choose between launching or delaying a new product
based on market conditions.
5. Forecasting Models – Techniques used to predict future events or trends based on
historical data and analysis.
Example: Forecasting next quarter’s sales using past sales data and seasonal patterns.
Academic Roots:
These methods are studied in fields like:
1. Operations Research – Focuses on applying mathematical models, optimization, and
analytical methods to make better business and operational decisions.

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

Example: Using linear programming to allocate limited resources efficiently.

2. Management Science – Emphasizes the use of quantitative analysis and modeling to

solve managerial and strategic decision-making problems.
Example: Applying forecasting models for production planning and inventory control.
3. Decision Support Systems (DSS) – Integrates data, models, and user-friendly
interfaces to assist managers in making informed decisions.
Example: A spreadsheet-based DSS that helps executives analyze investment options.
4. Simulation – Studies complex systems by imitating real-world processes to evaluate
outcomes under varying conditions.
Example: Simulating supply chain operations to identify bottlenecks and improve
efficiency.
Identification of the Problem and Environmental Analysis
No decision is made in a vacuum. It is important to analyze the scope of the domain and the
forces and dynamics of the environment.
Analyze:
Organizational culture
Decision hierarchy (centralized/decentralized)
Environmental factors and dynamics
This process is called Environmental Scanning and Analysis — monitoring and
interpreting relevant information.

BI/BA tools help identify potential problems.

Shared understanding among stakeholders is crucial for accurate modeling.
A) Variable Identification
Every model involves:
Decision Variables – controllable choices (e.g., amount to invest, number of employees)
Result Variables – outcomes to be optimized (e.g., profit, cost)
Uncontrollable Variables – external factors (e.g., interest rates, taxes)
Intermediate Variables – link decisions to outcomes (e.g., spoilage → affects profit)
Tools for variable identification:
Influence Diagrams – graphical models showing variable relationships.
Cognitive Maps – represent variables and interactions for complex decision problems.
B) Forecasting (Predictive Analytics)
Essential for prescriptive modeling since decisions affect the future.
Involves predicting:
Demand
Customer behavior

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

Market conditions
Online commerce has increased the need for rapid and accurate forecasting.
Forecasts support “what-if” and sensitivity analysis for future scenarios.

Model Categories
Decision models can be classified into seven groups (examples):
Model Type Explanation Example
Determine the best possible Finding the production
1. Optimization solution from many mix that maximizes profit
Models alternatives under given using Linear
constraints. Programming.
Imitate real-world systems to Simulating customer
2. Simulation
study their behavior under arrivals at a bank to
Models
different conditions. optimize service counters.
Use rules of thumb or Using the nearest
3. Heuristic approximation techniques to neighbor algorithm for a
Models find good (not always traveling salesman
optimal) solutions quickly. problem.
Using time-series
4. Forecasting Predict future values based
analysis to forecast
Models on historical data and trends.
monthly sales.
5. Decision Evaluate different decision Using a decision tree to
Analysis alternatives under choose between
Models uncertainty or risk. investment options.
Analyze waiting lines to Modeling airport check-
6. Queuing
optimize service efficiency in counters to minimize
Models
and resource utilization. passenger wait times.
Represent and optimize Finding the shortest
7. Network
systems of interconnected delivery route using
Models
nodes and paths. Dijkstra’s algorithm.

Models may be:

Static or Dynamic
Under Certainty, Uncertainty, or Risk
Tools Used:
Spreadsheets (Excel)
OLAP systems

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

Data mining tools

Modeling languages

Model Management
Models, like data, must be maintained and updated.
Managed using Model-Based Management Systems (MBMS) — analogous to DBMS for data.

Knowledge-Based Modeling
DSS (Decision Support Systems): Use quantitative models.
Expert Systems: Use qualitative, knowledge-based models.
Predictive analytics methods (classification, clustering) can enhance model knowledge.

Current Trends in Modeling

Trend Description
Model Libraries Reusable libraries and online repositories for models and solution
techniques.
Cloud-Based Tools Models and simulations run online; easy access and scalability.
Open Optimization Platforms e.g., NEOS Server (University of Wisconsin-Madison)
provides free optimization services.
Transparency Models increasingly hidden behind user-friendly dashboards and CRM
systems.
Parallel Computing & Data Warehouses Needed for large datasets and complex
model computations.
Influence Diagrams & Meta-Models Representing and analyzing “models of models” for
better understanding.
Example:
Revenue management models once limited to airlines are now used in retail,
entertainment, and insurance.

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

STRUCTURE OF MATHEMATICAL MODELS FOR DECISION

SUPPORT

Components of Mathematical Models

Component Description
Example
1. Result (Outcome) Variables Reflect system performance (dependent variables).
Profit, Market Share, Customer Satisfaction
2. Decision Variables Controllable by the decision maker.
Investment Amount, Pricing, Scheduling
3. Uncontrollable Variables External or environmental factors beyond control.

Interest Rates, Taxes, Fuel Prices

4. Intermediate Variables Link decision and result variables.
Spoilage, Employee Satisfaction

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

Structure (Mathematical Relationships)

Mathematical models use equations or inequalities to link variables.

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

CERTAINTY, UNCERTAINTY, AND RISK

The decision-making process involves evaluating and comparing alternatives. During this
process, it is necessary to predict the future outcome of each proposed alternative. Decision
situations are often classified based on what the decision maker knows (or believes) about the
forecasted results. We customarily classify this knowledge into three categories (see Figure 8.2),
ranging from complete knowledge to complete ignorance:
• Certainty
• Uncertainty
• Risk
When we develop models, any of these conditions can occur, and different kinds of
models are appropriate for each case.

Decision-Making under Certainty

All outcomes are known (deterministic environment).
Often applies to structured problems with short time horizons.
Simplifies modeling → enables optimal solutions.
Example: Investing in U.S. Treasury Bills with a fixed return.

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

Decision-Making under Uncertainty

Multiple possible outcomes, but probabilities are unknown.
Difficult to model due to lack of information.
Managers try to reduce uncertainty by gathering more data.
Example: Deciding to enter a new market with no historical data.

Decision-Making under Risk

Several outcomes with known or estimated probabilities.
Enables risk analysis — assessing the degree of calculated risk.
Most real-world business decisions fall into this category.
Expected Value Method used to select the best alternative.

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

DECISION MODELING WITH SPREADSHEETS

Model Development and Implementation
Models can be developed and implemented using various programming languages, modeling
systems, and software tools.
The focus is primarily on:
Spreadsheets (and their add-ins)
Modeling languages
Transparent data analysis tools
Spreadsheets are the most widely used tools because they are flexible, powerful, and user-
friendly.
Spreadsheets as Modeling Tools
Popularity and Strength
Spreadsheets are versatile implementation software used across business, engineering,
mathematics, and science.
They provide:
Extensive statistical, forecasting, and modeling functions.
Database management features.
Ease of use for end-users and decision makers.
Evolution of Spreadsheets
As spreadsheets evolved, add-ins were developed to extend their analytical and modeling
capabilities.
These add-ins help perform optimization, simulation, and intelligent analysis — vital for
Decision Support Systems (DSS).
DSS-Related Spreadsheet Add-ins

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

Many comparable add-ins are available for free or at low cost, and new ones are regularly
introduced in the marketplace.
Advantages of Using Spreadsheets for Modeling:

OLAP and Multidimensional Analysis:

OLAP (Online Analytical Processing) systems use multidimensional data cubes for
analysis.
Most OLAP systems resemble advanced spreadsheet interfaces, enabling:
Slicing and dicing of data.
Easy visualization of multidimensional data
Small data subsets can be imported into spreadsheets for ad hoc analysis.
Popular Spreadsheet Package: Microsoft Excel
Microsoft Excel is the most widely used spreadsheet tool for modeling and analytics.
Excel’s key features include:
Built-in Solver and Data Analysis ToolPak
Goal Seek and Scenario Manager
PivotTables for multidimensional analysis
Support for macros (VBA) and add-ins

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

Integration with databases and external data sources

Example: Loan Calculation Model
Static Model Example (Figure 8.3)

A loan calculation spreadsheet where:

Cells contain formulas (e.g., monthly payment, interest rate, loan amount).
A change in the interest rate (cell E7) immediately updates the monthly payment
(cell E13).
Useful for goal seeking (e.g., finding loan amount for a desired payment).
Static Model:
Represents a single point in time or one instance of calculation.
Example: A single month’s loan payment calculation.

Dynamic Model Example (Figure 8.4)

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

Models changes over time, showing the behavior of a system across multiple
periods.
Example: Loan repayment model showing effect of prepayment on principal
balance over time.
Dynamic Model:
Captures time-dependent behavior.
Common in forecasting, scheduling, and financial simulations.
Risk Analysis in Spreadsheets
Risk analysis can be incorporated through:
Built-in random number generators
Simulation models (e.g., Monte Carlo simulation using @RISK)
Allows decision makers to:
Evaluate outcomes under uncertainty.
Assess probability distributions of results.

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555
 ACHARYA INSTITUTE OF TECHNOLOGY
Affiliated to Visvesvaraya Technological University, Belagavi, Govt. of Karnataka.
Approved by AICTE, New Delhi and Accredited by NBA (AE, BT, CSE, ECE, ME and MTE)
Department of Artificial Intelligence & Machine Learning
and
Computer Science & Engineering (Data Science)

Key Spreadsheet Modeling Concepts

Acharya Dr. Sarvepalli Radhakrishnan Road, Soladevanahalli, ACHIT Nagar P. O., Bangalore-560 107
[Link] Ph.: 080 22555555