library / lib392311b2426e91c4
Operations Research for Social Good
In a sentence
A practical guide that teaches data scientists and operations research practitioners how to model and solve real-world social good problems using mathematical optimization with SAS and Python.
This book introduces the powerful world of Operations Research (OR) and mathematical optimization by applying it to real-life humanitarian and social challenges. Moving beyond theory, it provides a hands-on, practitioner's approach to formulating and solving complex decision-making problems. Through a series of compelling use cases—from optimizing food baskets for the UN World Food Programme to scheduling students during a pandemic and improving breast milk donation logistics—you'll learn the art of modeling. Each case is demonstrated with complete code in both SAS OPTMODEL and Python with Pyomo, making it an invaluable resource for data scientists, graduate students, and analytics professionals looking to apply their skills to make a tangible, positive impact on the world.
The four lenses
- Science
- Statistics
- Systems
- Strategy
The model
This model outlines the process by which applying operations research principles—specifically, clear problem formulation, appropriate model selection, and quality tool implementation—leads to higher quality solutions. These solutions, in turn, improve organizational decision-making, resulting in greater resource efficiency and enhanced social impact for mission-driven organizations.
Problem Formulation Claritydesign lever
The degree to which a real-world problem is successfully translated into a formal optimization structure, with clearly identified decision variables, constraints, and objective functions.
Appropriate Model Type Selectiondesign lever
The choice of the correct optimization model type (e.g., LP, MILP, NLP, Network) that accurately represents the mathematical characteristics of the problem, such as variable types and linearity of relationships.
Tool Implementation Qualitydesign lever
The accuracy and efficiency of translating the mathematical formulation into a specific programming language and modeling package, such as SAS OPTMODEL or Python/Pyomo.
Solution Qualityoutcome metric
The degree to which a computational solver can find a feasible, and ideally optimal or near-optimal, solution to the formulated problem within an acceptable timeframe.
Decision-Making Qualitybehavioral pattern
The extent to which an organization adopts the model's outputs to guide its actions and strategic choices, moving from heuristic-based to evidence-based decision-making.
Resource Efficiencyoutcome metric
The improvement in the organization's ability to achieve its goals with minimal waste of financial, human, or material resources, as a result of implementing optimized decisions.
Social Impactoutcome metric
The ultimate positive change achieved for the organization's target population or cause, such as improved health outcomes, wider access to services, greater safety, or reduced hunger.
How they connect
- problem formulation clarity → influences solution quality
- model type selection → influences solution quality
- tool implementation quality → influences solution quality
- solution quality → influences decision making quality
- decision making quality → influences resource efficiency
- decision making quality → influences social impact
- resource efficiency → influences social impact
The story
The reader The reader is a data scientist, analytics professional, operations research practitioner, or graduate-level student. They have technical and analytical skills but want to move beyond descriptive and predictive analytics into the realm of prescriptive analytics. They are motivated not just by technical challenges but also by a desire to apply their skills to 'Data4Good' initiatives and make a positive impact on society.
External problem
The reader needs to learn how to formulate and solve complex, real-world decision-making problems using mathematical optimization, but theoretical textbooks are too abstract and don't provide practical application or code.
Internal problem
The reader feels their analytical skills could be used for more meaningful work but feels intimidated by the perceived complexity of Operations Research. They are uncertain about how to translate a vague social problem into a precise mathematical model.
Philosophical problem
It's just plain wrong that powerful decision-making tools like mathematical optimization are mostly confined to corporate profit-making when they could be used to more efficiently allocate scarce resources for humanitarian aid, healthcare, and public services.
The plan
- Understand the Landscape: Grasp the fundamentals of mathematical optimization and its place within advanced analytics.
- Learn by Doing: Follow along with real-world use cases, progressing from Linear Programming to more complex models like Mixed Integer, Nonlinear, and Multicriteria Optimization.
- Master the Tools: For each use case, study the provided code in both SAS OPTMODEL and Python/Pyomo to learn how to translate a mathematical formulation into a working model.
Success
- The reader becomes a confident OR practitioner, capable of taking a complex social problem, formulating it as a mathematical model, and solving it using industry-standard tools.
- They can build prescriptive analytics engines that help non-profits and mission-driven organizations make better, more efficient decisions with their limited resources.
- Their career is enhanced with a scarce and valuable skill set, allowing them to lead high-impact 'Data4Good' projects.
At stake
- The reader remains stuck in the world of descriptive and predictive analytics, unable to provide actionable, optimal recommendations.
- Mission-driven organizations continue to make suboptimal decisions based on intuition or simple heuristics, wasting precious resources that could be used to help more people.
- The reader misses the opportunity to apply their analytical talents to solve some of the world's most pressing challenges.
Chapter by chapter
ch04Linear Programming
This chapter introduces Linear Programming (LP) as a pivotal optimization methodology, outlining its principles and applications through real-world case studies, particularly highlighting its use in humanitarian logistics.
ch06Nonlinear Programming
This chapter examines nonlinear programming (NLP) in optimization problems, presenting its principles and application in real-world scenarios, specifically in medical and environmental contexts.
ch07Network Optimization
This chapter explores the use of network optimization to address complex logistical challenges, illustrated through the Traveling Salesman Problem and Vehicle Routing Problem within the context of public health logistics.
- Network optimization is crucial for solving complex logistical problems, particularly in public health logistics.
- The Traveling Salesman Problem (TSP) can effectively model efficient routing challenges, minimizing travel distance while ensuring all required locations are visited.
- The introduction of demand constraints in the Vehicle Routing Problem (VRP) allows for more realistic modeling of logistics involving multiple vehicles and varying capacities.
- Addressing these optimization problems can significantly enhance the efficiency of healthcare delivery, impacting patient diagnosis and treatment.
ch08Multicriteria Optimization
This chapter explores the complexities of multicriteria optimization, demonstrating how multiple, often competing, objectives can be balanced using systematic approaches, with a real-world application in police patrol scheduling serving as a primary case study.
- Multicriteria optimization requires navigating trade-offs between competing objectives, epitomized by the challenge of scheduling police patrols.
- The Pareto frontier serves as a critical concept in assessing nondominated solutions in optimization problems.
- Effective scheduling is not only a logistical necessity but also central to ensuring community safety and trust in public services.
- Practical application of optimization techniques, as demonstrated through the SFPD use case, provides actionable strategies for decision-makers.
ch09Practice Problem Solutions
This chapter provides step-by-step solutions to a series of complex optimization problems drawn from various mathematical programming disciplines, equipping readers with practical applications of theoretical concepts.
Related in the library