Regression Trees to Model Decisions and Uncertain Factors in a Water Quality Decision Problem
Balancing water quality standards while facilitating economic growth with uncertain factors in a complex system is challenging for policy makers. This case study analyses the fictional town of Fortuna, the town manages two sources of pollution by adjusting the proportions of pollution treated at each source. Higher treatment proportions improve river water quality but also increase costs, leading to higher tax rates on residents.
This case study evaluates modelling relationships between the combination of decision variables and uncertain factors. There are 6 uncertain factors that influence water quality varying within a specified range. Regression trees are applied to evaluate system performance – using two water quality and two economic performance metrics.
Regression trees facilitated insights into the significance of uncertain factors and decision variables combinations on system performance. Additionally, they create highly interpretable framework allowing policy makers without specific systems knowledge to make informed decisions.
Ìý