variationalDCM: Variational Bayesian Estimation for Diagnostic Classification Models
Keiichiro Hijikata, Motonori Oka, Kazuhiro Yamaguchi, Kensuke Okada · 2023
Enables computationally efficient parameters-estimation by variational Bayesian methods for various diagnostic classification models (DCMs). DCMs are a class of discrete latent variable models for classifying respondents into latent classes that typically represent distinct combinations of skills they possess. Recently, to meet the growing need of large-scale diagnostic measurement in the field of educational, psychological, and psychiatric measurements, variational Bayesian inference has been developed as a computationally efficient alternative to the Markov chain Monte Carlo methods, e.g., Yamaguchi and Okada (2020a) , Yamaguchi and Okada (2020b) , Yamaguchi (2020) , Oka and Okada (2023) , and Yamaguchi and Martinez (2023) . To facilitate their applications, 'variationalDCM' is developed to provide a collection of recently-proposed variational Bayesian estimation methods for various DCMs.