Author(s): Leonardo Alfonso; Han Wang; Schalk Jan Van Andel
Linked Author(s):
Keywords: Decision making under uncertainty; Decision Trees; Prospect Theory; Early Warnings; Flood Forecasts; Simulation
Abstract: Decisions about early warnings are generally taken after a process that involves data collection, modelling and decision-making. A challenge is that each of these steps have some uncertainties associated, and decision-making becomes difficult. A question that arises is: to what extent it would be possible to automate decisions related to issuing flood early warnings? To answer this question, we propose to simulate decision-making outcomes using two knowledge communities, in which the related decision has an uncertain component regarding the occurrence of the flood event. First, the machine learning community, from which Decision Trees have been developed. Second, the behavioral economics community, from which Prospect Theory has studied decision making under uncertainty from the human perspective. Results show promising insights to automate decisions in flood early warning, in particular if both proposed models are combined.
Year: 2018