Distilling the Pareto Optimal Front into Actionable Insights

PyretoClustR is a modular framework tool, designed to make multi-objective optimization (MOO) results more accessible and easier to use in environmental decision-making. MOO often produces highly complex Pareto fronts, which can be overwhelming for both scientists and stakeholders who are trying to interpret trade-offs among competing objectives like agricultural productivity, biodiversity, water quality, and ecological flow.

PyretoClustR supports this process, while adapting to various environmental datasets and decision-making scenarios. It automatically selects effective parameters for principal component analysis, clustering, and outlier handling (managing unusual values, very different from others). PyretoClustR successfully reduced the Pareto front, a set of best possible solutions in a multi-objective optimization setting, from 2419 points to 18 representative solutions, facilitating understanding of MOO for informed decision making. Its purpose is to produce user-friendly visualizations of the selected dataset for decision-makers to be able to better understand the situation and make science-based decisions.

Link to the paper: Distilling the Pareto optimal front into actionable insights - ScienceDirect