Evaluating model complexity in fisheries management procedures

本研究的首要目标是评估渔业管理中模型复杂性的鲁棒性。在过去的几十年里，现代股票评估模型变得异常复杂。这种复杂性的增加似乎是合理的，因为生物真实性的增加和模型对更复杂的数据的拟合。简单模型与复杂模型之间存在着众所周知的偏差-方差权衡；从渔业管理的角度来看，模型复杂性的增加是合理的吗？还是随着模型复杂性的增加，我们正在失去预测能力？本研究旨在通过两个项目来解决这一问题：1)通过模拟估计实验和实际案例研究来评估简单模型与复杂模型在政策参数方面的鲁棒性；2)通过管理程序评估来设计更好的数据丰富和数据贫乏情况下的管理程序。管理程序评价是渔业科学中用于设计和检验渔业管理程序的最新工具；它建立在处理动态系统行为的系统控制理论的基础上。

The over-arching objective of this research is to evaluate the robustness of model complexity in fisheries management. Over the past few decades modern stock assessment models have grown incredibly complex. This increase in complexity appears to be justified by increasing biological realism and fitting models to more complex data. There is a well known bias-variance tradeoff associated with simple versus complex models; is this increase in model complexity justifiable from a fisheries management perspective or are we losing predictive power with increasing model complexity? The proposed research aims to address this question through two projects: 1) evaluate the robustness of simple versus complex models with respect to policy parameters using simulation estimation experiments and a real case-study, and 2) design better management procedures for data rich and data poor situations using management procedure evaluation. Management procedure evaluation is the newest tool being used in fisheries science for designing and testing fisheries management procedures; it is based on systems control theory that deals with the behaviour of dynamical systems.