Decision
AnalysisCornell

Certificate Program

Many real-world decision situations are complicated by uncertainty, complexity, and competing objectives. When you begin to frame and analyze a complex decision, you quickly realize: Defining the problem is the problem.

In this course, you will explore the concepts and tools you'll need for framing and analyzing complex decisions. You will define and frame the key components of a problem and identify the values that will inform your decision. You will then begin to map your values to your objectives with applicable examples. Finally, you will apply decision analysis to a wicked problem. Overall, you will examine how your unique perspective can alter how you frame and make complex decisions.

As you begin to frame your decisions, it's important to organize the problem in a manner that allows for the proper examination of solutions and their varying impacts. At this point, you can begin to incorporate the use of decision trees to meet that goal. What are the various kinds of decisions you can make? How is the impact of each decision path weighted and differently valued? Constructing decision trees will provide you with some answers to these questions.

In this course, you will explore the aspects of decision trees and when to best incorporate them into your decision analysis. You will map how performance measures may influence different sequences and outcomes. You will calculate risk profiles, develop modeling measures, and decide on the proper usage of decision trees in discrete choice situations.

While decision trees are commonly and widely used, they won't always be the right approach; you will explore in detail why that is and when it can make the most sense in given situations.

When you conduct your quantitative analysis, the fact that it is quantitative may seem to imply that the data is objectively set in a definitive manner and not suited for subjective analysis. In reality, quantifiable metrics include subjectivity at several points, including how the problem is approached and how you elicit the observations of knowledgeable individuals. It may surprise you, but you will often encounter subjectivity while you're trying to frame a decision process that is objective. Many times you'll discover that the questions that you're asking, the types of probabilities you need, and other information won't involve perfectly clear observations. Making a decision, therefore, isn't always as simple as picking the "best" quantitative option.

When you consider potential solutions, how does the elicitation of subjective data come into play? In this course, you will analyze the processes and theories you must consider as you begin to explore the subjectivity of objectivity in decision making and how they are related. You can potentially elicit subjective expertise to get a sense of the demand for a certain approach. You can also think about the overall riskiness of your potential choice as you deliberate over options. And there are biases and heuristics that may affect your decision. This all points to the importance of understanding the value of information and how it might influence your decisions.

Evaluating how the subjectivity of objectivity can affect how you view options will provide added context and tools for decision making. These elements will help guide you as you further consider how to frame your choices.

Making a decision often isn't as simple as choosing between Choice 1 and Choice 2 because one option is clearly better than the other. In the real world, not only are there complexities regarding financial costs, but there are deeper considerations around risk. Adding risk into the equation then requires considering whether the more profitable option is not worth the risk, and adding utility theory allows for the proper modeling of these complex choices. Risk and other considerations can subsequently be placed into a decision tree for further evaluation.

You have explored how subjective probabilities and the value of information can be integrated into your decision analysis. In this course, you will examine how risk attitudes and utility theory also impact your decision analysis. You will identify how risk attitudes are related to several axioms and paradoxes. You will also use utility theory to devise a model for quantifying subjective inputs to a decision then apply these additional considerations to a decision tree framework.

Risk and utility theories have the potential to totally change how you may have originally framed your decision, but this fuller picture provides invaluable insight into how you should make your decisions. It will also offer clearer context for when you incorporate more complicated analysis tools into your analysis process.

In complex engineered systems and design processes, you do not have the luxury of having fully defined decision alternatives with a clear mapping of their performance trade-offs or consequences. As you transition from classical discrete-choice problems, problem structuring and evaluation tools must become more advanced in their ability to explore large design problems while providing innovative decision analytical tools for helping to clearly map possibilities, their trade-offs, and key consequences.

In this course, you will discover how to formulate multi-objective design problems and more formally consider their trade-offs using the concepts of non-dominance, Pareto optimality, and a posteriori decision support.

As computational simulation becomes more commonplace in design and decision processes, it is important for you to consider how uncertainty could be impacting your perceptions of performance, trade-offs, and consequences. A single simulation can be seen as mapping from your design alternative to its performance objectives based on a single set of assumptions and choices used in the model's representation of the system of interest. Monte Carlo simulation can be thought of as accounting for uncertainties in your modeling assumptions and choices where you can simulate performance if your design resides in many different but plausible alternative worlds (i.e., many states of the world).

In this course, you will broaden the types of performance measures that can be used in your decision framings to include risks and vulnerabilities. You will assess the value of Monte Carlo simulation in better understanding the sensitivities, risks, and consequences of your candidate design alternatives. You will also explore the emerging insights and analytics associated with decision making under deep uncertainty. Given the many ways that our decisions shape concerns surrounding people, profit, and planet, finding solutions that maintain acceptable performance across many plausible futures then explicitly mapping their vulnerabilities is extremely valuable.

Decision making for complex systems often necessitates the modeling of system dynamics, optimization across multiple conflicting objectives, the analysis of uncertainty, and the visual analysis of performance. The depth of analysis is often limited by the tools available to decision makers. Recently, however, a number of software packages have been developed and deployed specifically to aid in decision making for complex systems.

In this course, you will explore the latest options in real-world decision making in the face of uncertainty. You will also use the open source Python library Rhodium to examine the shallow lake problem by testing multiple problem formulations, examining trade-offs between conflicting objectives, and discovering consequential combinations of uncertainty. In your final project, you will examine your decisions in simulated situations of uncertainty.