Using Diagnostic Classification Models to Validate Attribute Hierarchies and Evaluate Model Fit in Bayesian Networks
Using Diagnostic Classification Models to Validate Attribute Hierarchies and Evaluate Model Fit in Bayesian Networks
复制标题
DOI：
10.1080/00273171.2019.1632165
复制
发表时间：
2019-07-05
影响因子：
3.8
通讯作者：
Templin, Jonathan
中科院分区：
文献类型：
--
作者：
Hu, Bo;Templin, Jonathan
We investigate the relationship between Bayesian inference networks (BayesNets) and diagnostic classification models (DCMs). Specifically, we demonstrate and empirically examine the equivalency of parameterizations between BayesNets and DCMs. Then, we propose a model-comparison framework for testing the model fit of BayesNets, in which we show how BayesNets are nested within the saturated DCM structural models. Additionally, we show when attributes feature a linear hierarchy, the Hierarchical DCM is nested within both BayesNets and saturated DCMs. The usefulness of proposed framework and model-fit testing strategy was supported by the results of analyzing both simulated and empirical data.
