Skip to main content
Have a personal or library account? Click to login
Rhodium: Python Library for Many-Objective Robust Decision Making and Exploratory Modeling Cover

Rhodium: Python Library for Many-Objective Robust Decision Making and Exploratory Modeling

Open Access
|Jun 2020

Figures & Tables

Figure 1

The four steps of the many objective robust decision making (MORDM) framework, as applied using the Rhodium library. The process typically begins with problem formulation. Each step facilitates stakeholder collaboration using the generated visual analytics. Figure adapted from [12].

Figure 2

Core classes of the Rhodium library.

Figure 3

Non-linear dynamics of the irreversible lake model, with the phosphorus recycling in orange and phosphorus sink in black. The equilibria of this system are presented as points, with black denoting the stable equilibria and white denoting the unstable equilibrium, i.e., the tipping point of this system. Increasing the phosphorus concentration and crossing this tipping point puts the system at an irreversible eutrophic state.

Figure 4

Four alternative visualization options available in the Rhodium library. Each panel presents the performance of each candidate solution in: (a) a two-dimensional scatter plot with the color of each point set by the performance of each solution on one of the objectives; (b) a two-dimensional scatter plot with the color of each point set by whether it meets user-set preference criteria; (c) a three-dimensional scatter plot with the size of each point set by the performance of each solution on one of the objectives; (d) a parallel axis plot, where each objective is represented by a vertical axis and the performance of each solution (each line) is indicated by the point where it crosses each axis.

Figure 5

Scenario discovery results as produced by PRIM and Cart, using the Rhodium library. (a) One of the identified PRIM boxes, representing parameter ranges where the policy chosen is always reliable. The “Prev” and “Next” buttons allow the user to navigate to other PRIM boxes with different coverage and density of reliable SOWs. (b) The classification tree produced by Cart. Each node in the tree represents an orthogonal split on the range of one of the parameters. Moving downward on the tree indicates additional divisions of the parametric space.

DOI: https://doi.org/10.5334/jors.293 | Journal eISSN: 2049-9647
Language: English
Submitted on: Aug 23, 2019
Accepted on: Apr 30, 2020
Published on: Jun 9, 2020
Published by: Ubiquity Press
In partnership with: Paradigm Publishing Services
Publication frequency: 1 issue per year

© 2020 Antonia Hadjimichael, David Gold, David Hadka, Patrick Reed, published by Ubiquity Press
This work is licensed under the Creative Commons Attribution 4.0 License.