论文部分内容阅读
在渔业资源评估中,CPUE(catch per unit effort)标准化是基础性工作。一般线性模型(generalized linear model,GLM)已成为CPUE标准化的基本方法,但GLM模型在误差结构、自变量的选择、缺失数据、复杂交互效应及异常值处理等方面仍然缺乏灵活性。本文基于模拟数据及我国东、黄海鲐鱼(Scomber japonicus)灯光围网渔业数据,比较和分析了基于GLM模型与回归树模型在CPUE标准化中的效果。研究表明:当渔业数据不存在非线性关系与异常值时,GLM模型与回归树模型均能较好地对CPUE进行标准化,但由于回归树模型具有阶跃函数特征,因而GLM模型更具优势;在非线性关系及异常值存在的条件下,回归树模型对CPUE的标准化具有相对较小的估计误差,模型更简约、有效。由于回归树模型能可视化显示自变量与应变量间的复杂关系,因此,更有利于探索和分析渔业数据。
In the fisheries resources assessment, CPUE (catch per unit effort) standardization is the basic work. The generalized linear model (GLM) has become the basic method of CPUE standardization. However, the GLM model lacks flexibility in terms of error structure, selection of independent variables, missing data, complex interaction effects and outlier processing. Based on the simulated data and the light seine fishery data of Scomber japonicus in China, the effects of GLM model and regression tree model in the standardization of CPUE were compared and analyzed. The results show that GLM model and regression tree model can both normalize CPUE when there is no nonlinear relationship and abnormal value in fishery data, but GLM model has more advantages because of the step function of regression tree model. Under the condition of nonlinear relationship and outliers, the regression tree model has a relatively small estimation error for the standardization of CPUE, and the model is more simple and effective. Since the regression tree model can visualize the complex relationship between the independent variables and the dependent variables, it is more conducive to the exploration and analysis of fishery data.