%0 Journal Article %T 基于Fiducial预测密度的模型选择
Model Selection Based on Fiducial Predictive Distribution %A 范晨 %J Advances in Applied Mathematics %P 192-202 %@ 2324-8009 %D 2025 %I Hans Publishing %R 10.12677/aam.2025.143105 %X 模型选择作为统计分析的一个重要工具,它能选出候选模型集中拟合数据生成过程最好的那个模型。为了得到更好的预测标准,本文提出了一种基于Fiducial预测密度的模型选择方法,同时加入容许集来进一步压缩候选模型集,给出理论的同时也给出了MH算法去应用它。最后,对本文提出的模型选择进行了模拟研究与实例分析,结果均表明我们的方法优于其他方法。
Model selection is an essential tool in statistical analysis, enabling the identification of the best-fitting model for the data-generating process from a set of candidate models. To achieve better predictive performance, this paper proposes a model selection method based on Fiducial predictive density, incorporating an admissible set to further reduce the candidate model space. Theoretical foundations are provided, along with a Metropolis-Hastings (MH) algorithm for its implementation. Finally, simulation studies and empirical analyses are conducted to evaluate the proposed method. The results demonstrate that our method outperforms existing approaches in terms of both predictive accuracy and efficiency. %K 模型选择, %K 容许集, %K Fiducial预测密度
Model Selection %K Admissible Set %K Fiducial Predictive Distribution %U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=109423