Decision and Risk Modeling
90-775
Units: 6
Description
This course is one of several introductory courses in analytics and management science that survey a variety of hands-on quantitative modeling methods that are useful to decision makers and business and policy analysts. They overlap considerably, but are tailored for particular programs based on student career paths and interests.
- MSPPM, MS3, and PPM-DC students take two minis: Management Science I and II (90-722, 90-760) in spring of Year 1.
- PPM-DA & HCAIT students take a two-semester sequence:
- Optimization (90-755) and Decision and Risk Modeling (90-775) in spring of Year 1.
- Decision Analytics for Business and Policy (94-867) in Year 2.
- MISM students take one mini: Decision Making Under Uncertainty (95-760).
Students in all three of those tracks, as well as the Army-BIDA program, are invited to take 94-833 Multicriteria Decision Making & Decision Analysis as an elective, which covers a different but complementary set of managerial decision tools.
Learning Outcomes
The 90-755/90-775/94-867 sequence has six objectives.
First, you should learn about a variety of management science techniques, what they are capable of, and what their limitations are so that you can intelligently call upon specialists and consultants when the occasion arises. (90-755/90-775)
Second, you should become facile at building and analyzing models in both Excel (90-755/90-775) and advanced, state-of-the-art software that can handle larger problem instances (94-867).
Third, you should develop skill at writing mathematical equations that represent the “physics” (operational reality) of systems that you will be responsible for managing and improving. (All courses in the sequence).
Fourth, you should understand the inner workings of the algorithms that computers use to solve these equations (“models”) well enough to know which types of problems are “hard” or “easy” for the computer to solve, what that implies for the “size” problem that can be handled, and some tricks for transforming “hard” into “easier” problems to solve. (90-755 & 94-867)
Fifth, you should develop sophistication in your ability to interpret model solutions’ implications for practical managerial decision making. That will be done in 90-755/90-775 via small cases that help you connect the methods to problems that are richer than the typical end of chapter problem, and in 94-867 via larger cases and a term project.
Sixth, you should develop skill at addressing complex, ill-defined, large-scale decision-making problems that arise in policy and business. This will be done by learning the theory and analytical toolsets for executing a “data-to-decisions” pipeline in a principled way. In 90-867, students build on the management science methods learned from 90-755 and 90-775, along with tools from other Heinz courses (e.g., predictive tools from machine learning and statistics), to convert data to information (using regression, neural networks, hybrid mechanistic and ML models) and to embed that information in optimization models (via contextual optimization, decision-making under uncertainty and risk, and iterative learning-and-optimization methods such as approximation dynamic optimization and reinforcement learning). Students will work in Python and utilize a set of advanced data-to-decision packages (Gurobi, GurobiML, sklearn) to solve large cases.
Prerequisites Description
College pre-calculus and 90-755 or 90-722, including fluency graphing and interpreting linear and nonlinear functions.
Ability to work fluently with concepts from probability including the Binomial and Normal random variables, distributions more generally, computation of mean & standard deviation, event probabilities, and Bayes Rule. Some concepts in 90-775 are easier to grasp if one knows regression, but regression per se is not a prerequisite.
The course uses Excel intensively. If you have not become proficient with Excel in some other way, you should work through some Excel tutorials before the course begins.