Taking Decision Trees to the Next Level: Adding Uncertainty

March 02, 2021 | 07:00:AM - 07:00:PM CT

  • Making the Most of Data Mining and Visualization

  • On-Demand Videos


Speaker Profile Picture

Steve Kramer, Ph.D., MBB

Associate Professor of Decision Sciences

Nova Southeastern University

We start with a brief review of decision trees: the formulation of the logic tree based on the problem. We then explain (review) the L to R analysis in computing EMV. We then show through an example how this approach is applied: the case study is based on a paper the author published previously on decision-making under conditions of complex demand and market risks in the context of a healthcare resource capacity decision for a large urban hospital. The case considered is time-sensitive in that a cardiac unit was considering expansion based on expected demand if deregulation occurred in cardiac services within that state. Decisions had to be made in the near term for capacity to be in place if deregulation happened at an unknown point in time. There are many factors at stake: demand for open-heart surgery, hospital market share, etc. Investing too soon will result in over-capacity, while delaying too long could mean that the hospital could not meet market demand. We show the model with and without the Monte Carlo-applied distributions to convey the difference in insights from the approaches. CLICK HERE to complete the session survey