Course No 1503550

Course Name Optimization and Decision Models

Semester Spring 2023/2024

Report Date 08-04-2024

Team Members

U22200002 Asma Saif Almazrouei

U22200823 Hajer Eisa Almazmi

Table of Contents
Title Page
1. Introduction 3

2. Use Case 1: Efficient Human Resource Allocation for Amazon

o Background and operational challenges at Amazon. 4

o Objectives of using linear programming in staffing strategy. 4

o Methodology: Model formulation and constraints. 5

o Methodology: Optimization Application (MATLAB) 5

o Results, Discussion, and Future Work 6

o Strategic Impact and Conclusion 7

3. Use Case 2: Optimizing Interview Scheduling for Noon's Market Research

Strategy

o Background of Noon and the need for market research. 8

o Objectives of optimizing interview scheduling. 8

o Methodology: Model formulation and constraints. 8

o Methodology: Optimization Application (MATLAB) 9

o Results and discussion, and Future Work 10

o Strategic Impact and Conclusion. 10

4. Summaries of lessons learned, and personal experiences and knowledge

11
gained

5. Responsibilities 12

6. Appendices 13
1. Introduction

In the dynamic realm of e-commerce and technology, companies continually seek innovative
strategies to optimize their operations, enhance efficiency, and maintain a competitive edge. Linear
programming, a mathematical modeling technique, emerges as a powerful tool in this quest, offering
a structured method to make the best possible decisions within given constraints. This project delves
into two distinct use cases from leading e-commerce platforms, illustrating the versatility and
effectiveness of linear programming in addressing complex operational challenges.

The first case study explores Amazon's implementation of linear programming to refine its human
resource allocation strategy. As Amazon handles the difficulties of managing a vast and diverse
workforce across various operational shifts, linear programming aids in developing an optimal staffing
model. This model aligns workforce distribution with operational demands, ensuring cost efficiency
while sustaining service excellence and adaptability in a fluctuating market landscape.

The second case pivots to Noon, a leading e-commerce entity in the Middle East, utilizing linear
programming not only to update its market research interview scheduling but also to present a
graphical representation of the optimization data. This approach enables Noon to efficiently allocate
resources for conducting interviews, ensuring a representative sample of its customer base, and
observing to budgetary constraints. The graphical representation provides a clear, visual
comprehension of the optimization results, enhancing the strategic planning and decision-making
process.

Through these cases, the project aims to showcase the application of linear programming in diverse
operational contexts—ranging from workforce management to market research optimization. Each
case underscores the method's capacity to facilitate data-driven decisions, optimize resource
allocation, and strategically navigate the complexities inherent in the e-commerce and technology
sectors.
2. Use Case 1: Efficient Human Resource Allocation for Amazon
o Background and Operational Challenges at Amazon
Amazon, a powerhouse in both online shopping and technology, started as a groundbreaking online bookstore and
has since expanded into a massive e-commerce platform. This growth has solidified Amazon's position as a leading
force in the tech industry, known for its innovative contributions and its influence on changing how people shop
and what they expect from retailers. Driven by a desire to constantly push the boundaries, Amazon has developed
a wide range of products and services, which requires a complex and well-organized system to manage its rapid
expansion and the ever-changing demands of the marketplace.
Amazon's operation system is complex, especially in managing its workforce and allocating resources effectively.
As it transformed from a simple online retailer to a global technology powerhouse, Amazon developed a nimble
and sophisticated approach to handle its diverse and ever-changing workforce needs. The company's expansion and
broadening of services have made it challenging to efficiently allocate human resources across various shifts. Given
the fluctuating nature of its operations, which include e-commerce and logistics, Amazon must ensure accurate
staffing to meet the demands at any given time, whether during normal business hours, high-demand seasons, or
weekends. Finding the right balance between operational efficiency and cost-effectiveness is critical for Amazon
to stay ahead in the competitive technology industry while continuing to innovate and expand.

o Objectives of Using Linear Programming in Staffing Strategy

This case study explores Amazon's approach to optimizing human resource allocation, a critical aspect of
maintaining its leading position in the fast-paced e-commerce and technology sector. It centers on formulating a
staffing strategy that optimizes operational efficiency and cost-effectiveness while being flexible to market shifts.

Using linear programming, the study proposes a shift structure tailored to Amazon’s requirements across different
operational periods, aiming to cut labor costs without compromising service delivery. The strategy involves
mathematical modeling to handle workforce management challenges and to ensure responsiveness to consumer
demands within budget.

The analysis aims to highlight linear programming as a key tool for devising effective staffing strategies in
Amazon’s complex operations, emphasizing the importance of data-driven decisions in improving efficiency and
sustaining a competitive advantage.
o Methodology: Model Formulation and Constraints
The methodology adopts linear programming to create an optimization model aimed at finding the most cost-
effective staffing distributions. This model is comprised of several key components:
• Objective Function: The goal is to minimize the total labor cost incurred across different shift periods,
including regular, peak, and weekend times. This function quantifies the primary aim of reducing expenses
while maintaining operational efficiency over the year.
• Constraints: These are set to ensure that the model's solutions are viable within Amazon's operational
framework. Constraints cover a range of requirements, such as the minimum number of staff needed for
each shift type, the specific costs associated with individual shifts, and an overall cap on the total weekly
labor expenditure. These limitations are crucial for ensuring that the staffing levels are both sufficient for
Amazon's needs and in line with its financial strategies.
Amazon is enhancing its staff scheduling strategies to enhance cost efficiency by aligning workforce allocation
with specific shift requirements. The approach involves an optimization model built on linear programming to
minimize the total labor cost of conducting shifts. The model employs an objective function and corresponds to the
number of each shift type as shown in Table 1.
o Methodology: Optimization Application (MATLAB)

This optimization model is implemented using MATLAB’s 'linprog' function. 'Linprog' is adept at solving linear
programming problems by specifying the coefficients of the objective function and incorporating the necessary
constraints. These constraints mirror Amazon's operational requisites, including the minimum number of employees
required for each shift type and the costs tied to each shift category. Through this approach, the model adeptly
identifies staffing allocations that minimize costs without compromising on the capability to meet Amazon's diverse
and dynamic operational demands.

MATLAB Code explanation:

1. Define the objective function Constraints:

c_reg = 5000;
c_peak = 4000;
c_weekend = 7000;
C = [c_reg, c_peak, c_weekend];
A = [ -1, 0, 0; 0, -1, 0; 0, 0, -1 ,1, 0, 0; 0, 1, 0; 0, 0, 1; 7, 7, 2; ];
B = [-100; -150; -200,5000; 4000; 7000; 30000];
lb = [0; 0; 0];

2. Solve the linear programming problem using the 'linprog' function:

options = optimoptions('linprog','Algorithm','dual-simplex');
[x, fval] = linprog(C, A, b, [ ], [ ], lb, [ ], options);
 3. Display the result in a tabular format:
if isempty(x)
disp('No solution was found.’);
else shiftTypes = {'Regular Shifts'; 'Peak Hour Shifts'; 'Weekend Shifts’};
optimalNumberOfShifts = [x(1); x(2); x(3)];
resultTable = table(shiftTypes, optimalNumberOfShifts, ...
‘VariableNames', {'Shift Type', 'Optimal Number of Shifts'});
disp(resultTable);
fprintf('Minimum Total Cost for all Shifts: $%.2f\n', fval); end

o Results, Discussion, and Future Work: Operational Effectiveness and Cost

Reduction Insights

The results from the linear programming model provide a strategic staffing allocation for Amazon's operations
over the course of a year. The optimal number of shifts determined by the model is 100 for Regular Shifts, 150
for Peak Hour Shifts, and 200 for Weekend Shifts. These allocations are designed to align with the objective
of minimizing annual labor costs, all while satisfying the staffing requirements laid out in the constraints. The
model has adeptly determined a cost-effective staffing plan within these operational parameters, locating the
minimum total annual cost for all shifts at $2,500,000.
In discussing these findings, it's evident that Amazon can maintain operational efficiency and cater to customer
demand throughout the year without exceeding its budgetary limits by adhering to this shift schedule. This
strategic allocation effectively supports staffing during the busier peak and weekend times, which are typically
known for increased customer activity and demand. The imposed non-negativity constraint ensures that every
shift type is properly staffed, thus creating a balance between cost savings, operational efficacy, and the
flexibility required to adjust to shifts in market demand.
Looking to the future, it is imperative to enhance the model by incorporating additional constraints that address
the complexities of shift management. These enhancements include:
▪ Shift Change Time: Introducing constraints to manage the duration and timing of shift changes,
ensuring smooth transitions and adherence to labor regulations.
▪ Overtime Cost: Factoring in the additional costs associated with overtime work, which could
significantly impact the total labor expenditure.
▪ Night Shift Considerations: Including specific constraints for night shifts, which often entail
differential pay rates and unique staffing requirements.
Enhancing the model to include factors like shift change time, overtime costs, and night shift policies will
make it more robust and reflective of Amazon's real-world operations. A holistic approach will be fostered,
considering employee welfare, service quality, and labor regulations, leading to a sustainable staffing strategy
for Amazon. The adaptability of the model is critical to address demand fluctuations and cost changes while
ensuring that employee satisfaction and service quality are not compromised. This approach provides Amazon
with a foundation for effective human resource management, adaptable to evolving business needs.
o Strategic Impact and Conclusion

The application of linear programming to Amazon's staffing optimization has a multi-dimensional strategic
impact, which contributes significantly to the company’s operational resilience and competitiveness:
▪ Cost Efficiency: The model's core objective to minimize total labor costs is pivotal for maintaining
Amazon's low-cost leadership in the market. It directly affects the bottom line, enabling the company
to invest savings into other strategic areas like innovation or price competitiveness.
▪ Resource Allocation: Optimal resource allocation ensures that Amazon can meet the varied demands
of its operations without waste, utilizing human resources where they are most needed, thereby
improving overall service delivery and efficiency.
▪ Scalability: The model's design allows for adjustments in staffing that are scalable with Amazon's
growth and seasonal demand shifts. This ensures that as the company expands, its staffing strategy can
adapt without the need for continuous overhaul.
▪ Flexibility in Planning: Flexibility is inherent in the model, permitting rapid adjustment to staffing
plans in reaction to changes in market conditions or consumer behavior. This adaptability is critical for
sustaining operational continuity and for capitalizing on unforeseen opportunities.
▪ Employee Satisfaction: Although not explicitly modeled in the current staffing optimization,
employee satisfaction is an indirect beneficiary. A well-planned staffing strategy can lead to more
predictable shifts and potentially fewer instances of involuntary overtime, contributing to a better work-
life balance.
In conclusion, Amazon has successfully implemented a linear programming model to fine-tune its staffing
strategy, achieving an optimal balance between cost minimization and operational efficiency. This model,
powered by MATLAB’s, has allowed Amazon to make data-driven decisions about how many employees to
schedule for regular, peak, and weekend shifts, ensuring that each shift is adequately staffed according to the
company’s needs while keeping labor costs under control. The outcome of this model offers significant strategic
value, enhancing Amazon’s ability to adapt to changing demands and maintaining its competitive edge in the
market. Looking ahead, further refinements to the model are expected to bolster its effectiveness and contribute
positively to both the company’s bottom line and workforce management.
3. Use Case 2: Optimizing Interview Scheduling for Noon's Market
Research Strategy
o Background of Noon and the need for market research
Noon is a leading e-commerce platform in the Middle East, known for its wide range of products and services.
Founded with the vision to offer a comprehensive online shopping experience, Noon has quickly become a key
player in the region's digital economy. To sustain its growth and continue meeting customer needs effectively, Noon
recognizes the importance of in-depth market research. This initiative aims at refining Noon's approach to
conducting personal interviews, which are crucial for gathering rich consumer insights. These insights are
instrumental in shaping product development, enhancing customer experience, and informing strategic decisions.
Optimizing the scheduling of these interviews is crucial for ensuring broad and representative participation while
managing the logistical and financial aspects of market research. This optimization not only aids in effectively
gathering actionable insights but also aligns with Noon's commitment to being a customer-centric organization,
continually adapting to meet and exceed consumer expectations in a competitive marketplace.
o Objectives for optimizing interview scheduling
Noon prioritizes personal interviews in its market research to gain deep insights into customer experiences and
preferences, shaping the company's strategy. These insights drive product and service improvements, ensuring
alignment with consumer needs and supporting Noon's position in the e-commerce market. Allocating a budget for
these interviews highlights Noon's dedication to market understanding and innovation for customer satisfaction.
Careful interview scheduling to capture a representative customer sample underpins the research's relevance and
effectiveness, demonstrating Noon's commitment to a customer-oriented approach that responds to evolving
consumer needs and strengthens its market presence.
o Methodology: Model formulation and constraints.
The methodology Noon adopts for optimizing its interview scheduling employs linear programming to construct
an optimization model that is focused on minimizing the overall costs associated with conducting market research
interviews. This model is expressed with the following components:
o Objective Function: The aim is to reduce the total cost of carrying out interviews. This function captures
the main objective of cutting expenses while ensuring that the interviews are comprehensive and aligned
with the company's market research goals.
o Constraints: The model includes several constraints to ensure that the solutions are practical and feasible
within Noon's operational and budgetary limits. The constraints establish the minimum and maximum
number of interviews with families and single individuals, enforce the costs per interview type, and ensure
that all interview counts remain positive. These conditions are crucial for guaranteeing that the market
research efforts are effective and within the allocated budget as shown in table 2.
Noon can strategically schedule interviews to yield the most valuable market insights at the lowest possible cost.
This approach not only streamlines the process of gathering crucial consumer data but also ensures that the
company's resources are utilized in the most efficient manner possible, supporting Noon's overall strategic
objectives and enhancing its position in the competitive e-commerce landscape.
 o Methodology: Optimization Application (MATLAB)
In Noon’s market research optimization, MATLAB’s 'linprog' function is used for linear programming to minimize
the cost of scheduling interviews. It efficiently calculates the optimal number of interviews to conduct within
budgetary and participant representation constraints, ensuring Noon’s resources are utilized effectively.
1. Define the objective function Constraints

c_f = 1100;
c_s = 800;
C = [c_f, c_s];
A = [-1, 0;0, -1;1, 0;0,1;];
B = [-200;-70;400;100];
lb = [0; 0];

2. Solve the linear programming problem using the 'linprog' function

options = optimoptions('linprog','Algorithm','dual-simplex');
[x, fval] = linprog(C, A, b, [], [], lb, [], options);

3. Display the result in tabular format

if isempty(x)
disp('No solution was found.’);
else
householdTypes = {'Families '; 'Single Person’};
optimalNumberOfInterviews = [x(1); x(2)];
costPerInterview = [c_f; c_s];
resultTable = table(householdTypes, optimalNumberOfInterviews, costPerInterview, ...

'VariableNames', {'Household Type', 'Optimal Number of Interviews', 'Cost per Interview'});

disp(resultTable);

fprintf('Minimum Total Cost for all Interviews: $%.2f\n', fval);end

4. Graphical solution
The MATLAB code referenced in Table 3 utilizes the ‘linprog’ function for solving linear programming problems
with specified costs and constraints, yielding decision variables and the minimal cost for optimal interview
allocation.
o Results and discussion, and Future Work: Enhanced representation and
cost efficiency
The optimization of Noon's market research interview scheduling has proven to be effective, resulting in a minimum
total cost of $276,000. The strategy to conduct 200 family interviews and 70 single-person interviews reflects
Noon's commitment to a customer-centric approach, ensuring that diverse market segments are represented.
To enhance the model's accuracy, future modifications will integrate a variety of additional variables, such as:
▪ Precise Interview Timings: Aligning interview schedules with respondents' availability to increase
participation and data quality.
▪ Number of Children in Each Household: Tailoring questions and discussions to capture the impact
of family dynamics on consumer behavior.
▪ Participants' Income Brackets: Differentiating spending habits and product needs across various
economic segments.
These factors will enrich Noon's consumer insights, allowing for even more tailored market strategies.
Graphical methods have provided a foundational understanding of the model's outcomes. However, they have their
limitations, chiefly the ability to represent only two-dimensional data. Real-world scenarios require considering
more complex multi-dimensional data. Subsequent models will therefore utilize advanced computational techniques
capable of processing multiple variables simultaneously, transcending the constraints of two-dimensional graphical
representation and leveraging the full power of numerical optimization methods like those available in MATLAB.

o Strategic Impact and Conclusion

Noon’s optimized market research interview scheduling exemplifies their commitment to operational excellence
and strategic vision. The utilization of resources is not just improved for budget efficiency but also for the quality
of decision-making. Adopting linear programming to refine this scheduling process highlights Noon’s innovation
and pursuit of high-quality data, enhancing their strategic and operational agility.
The strategic impact of optimizing Noon’s market research interview scheduling is multifaceted and significant:
▪ Resource Allocation Efficiency: By minimizing costs and ensuring representative sampling, Noon
maximizes its market research efficacy, allocating resources where they are most needed.
▪ Budget Adherence: The optimization process ensures that market research activities stay within
financial constraints, demonstrating fiscal responsibility and strategic spending.
▪ Meeting Targets: Accurate scheduling aligns with target demographics and desired respondent
numbers, ensuring the research meets all predefined targets.
▪ Operational Planning: The application of linear programming in scheduling aids in the systematic
planning of operations, enhancing overall efficiency.
▪ Strategic Decision-Making: The insights gained through this optimized approach to market research
inform Noon’s strategic decisions, enabling data-driven actions that are likely to result in increased
customer satisfaction and business growth.
4. Summaries of lessons learned and personal experiences and
knowledge gained
Reflecting on the detailed exploration of linear programming applications within the e-commerce and technology
sectors, several key lessons and insights have emerged from the study of Amazon and Noon's use cases. These
insights not only underscore the practical utility of linear programming in solving real-world operational problems
but also highlight broader themes relevant to strategic management and operational optimization in rapidly evolving
industries.
Summaries of Lessons Learned
o Flexibility and Adaptability Are Crucial: Both Amazon and Noon demonstrated the importance of developing
flexible operational models that can quickly adapt to changing market demands and internal constraints. Linear
programming serves as a pivotal tool in this context, offering a structured yet flexible framework to reassess and
realign operational strategies as conditions evolve.
o Data-Driven Decision Making Enhances Operational Efficiency: The application of linear programming
underscores the value of data-driven decision-making processes. By quantitatively assessing various operational
scenarios and constraints, organizations can identify optimal strategies that balance cost, efficiency, and resource
allocation, leading to more informed and effective decisions.
o Strategic Resource Allocation Maximizes Competitive Advantage: Effective resource allocation, whether in
staffing or market research efforts, is fundamental to maintaining a competitive edge. Linear programming helps
pinpoint the most efficient distribution of resources, ensuring that companies like Amazon and Noon can operate
at peak efficiency while minimizing unnecessary expenditures.
o The Power of Visualization in Strategic Planning: The graphical representation of optimization data, as seen
in Noon's case, emphasizes the strategic value of visualizing complex data sets. Graphical methods can make
the outcomes of optimization models more accessible and logical, facilitating clearer strategic planning and
communication among stakeholders.
Personal Experiences and Knowledge Gained
o Applicability Across Various Operational Challenges: One of the most notable insights from this analysis is
the broad applicability of linear programming. It's attractive to see how a single mathematical technique can
address diverse challenges, from workforce management to scheduling market research interviews, highlighting
its fluctuation as a problem-solving tool.
o The Importance of Constraints in Planning: Understanding and defining the constraints within which
decisions must be made is as crucial as the decision-making process itself. This experience has deepened my
appreciation for the role of constraints in shaping strategic choices and the outcomes of optimization models.
o The Role of Technology in Operational Optimization: Exploring the use of MATLAB and linear
programming in these case studies has highlighted the critical role of technology and software tools in facilitating
operational optimization. It's a reminder of the continuous need for technological literacy and the integration of
software tools in strategic management practices.
In conclusion, the exploration of linear programming applications within Amazon and Noon's operational strategies
has been immensely enlightening, offering valuable lessons on flexibility, efficiency, and the strategic allocation of
resources. These insights not only enrich my understanding of operational optimization but also underscore the
importance of adopting an interdisciplinary and technologically informed approach to problem-solving in today's
complex business environment.
Responsibilities

Responsibility Team Member

1. Investigate several topics and select a suitable topic Asma and Hajer

2. Define the objective functions and constraints applicable to the

Asma and Hajer
chosen topic.

3. Execute the case studies utilizing MATLAB Asma and Hajer

4. Manually solve the cases and define the solution Asma and Hajer

5. Conduct testing on MATLAB to verify the accuracy of the solution Asma and Hajer

6. Prepare the presentation Asma and Hajer

7. Prepare and organize a comprehensive report of the project Asma and Hajer
Appendices
Table 1:
Minimize the total labor cost of conducting the shifts,
Objective 𝑪 = 𝑪_𝒓𝒆𝒈 × 𝒙_𝒓𝒆𝒈 + 𝑪_𝒑𝒆𝒂𝒌 × 𝒙_𝒑𝒆𝒂𝒌 + 𝑪_𝒘𝒆𝒆𝒌𝒆𝒏𝒅 × 𝒙_𝒘𝒆𝒆𝒌𝒆𝒏𝒅

𝑪_𝒓𝒆𝒈 Cost per Regular Shift

𝒙_𝒓𝒆𝒈 Number of Regular Shifts

𝑪_𝒑𝒆𝒂𝒌 Cost per Peak Hour Shift

𝒙_𝒑𝒆𝒂𝒌 Number of Peak Hour Shifts

Cost per Weekend Shift

𝑪_𝒘𝒆𝒆𝒌𝒆𝒏𝒅

Number of Weekend Shifts

𝒙_𝒘𝒆𝒆𝒌𝒆𝒏𝒅

Minimum of 100 staff per regular shift (x_reg ≥ 100)

Regular shift cost is $5000 (C_reg = 5000)

At least 150 staff during peak hours (x_peak ≥ 150)

Peak hour shift cost is $4000 (C_peak = 4000)

Constraints At least 200 staff during weekend days (x__weekend ≥ 200)

Weekend shift cost is $7000 per day (C_weekend = 7000)

Total weekly labor cost not to exceed $30,000 (7×C_reg×x_reg + 7×C_peak×x_peak +

2×C_weekend×x_weekend ≤ 30000)*

Non-negativity (x_reg, x_peak, x_weekend ≥ 0)

*7 stands for 7 days and 2 days for weekends.

Table 2:
Minimize the total cost of conducting the interviews,
Objective 𝑪 = 𝒄𝒇 × 𝒙𝒇 + 𝒄𝒔 × 𝒙𝒔

𝒄𝒇 Cost per Family Interview

𝒙𝒇 Number of Family Interviews

Cost per Single Person Interview

𝒄𝒔

𝒙𝒔 Number of Single Person Interviews

At least 50% of interviews with families (𝑥𝑓 ≥200)

At least 70% of interviews with a single person (𝑥𝑠 ≥70)

Total interviews with families not to exceed 400 (𝑥𝑓 ≤400)

Total interviews with single person not to exceed 100 (𝑥𝑠 ≤100)
Constraints

Non-negativity (𝑥𝑓 , 𝑥𝑠 ≥ 0 )

Total Cost = 1100 per family

Total Cost = 800 per single person

Table 3:

Component Description

linprog MATLAB function used to solve linear programming problems. subject to linear constraints.

Coefficients of the linear objective function. In the context of this code, C represents the costs per
f interview type,

These define the constraints related to the minimum and maximum allowable interviews.
A, b

lb lb ensures non-negative values for the number of interviews.

Specifies options for the linprog solver, such as the optimization algorithm.
options

x Decision variables that result from solving the optimization problem.

The minimum value of the objective function, given the optimal solution. Represents the total cost
fval of conducting the interviews at the optimal allocation.