variationalDCM (Q5980299)

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) <doi:10.1007/s11336-020-09739-w>, Yamaguchi and Okada (2020b) <doi:10.3102/1076998620911934>, Yamaguchi (2020) <doi:10.1007/s41237-020-00104-w>, Oka and Okada (2023) <doi:10.1007/s11336-022-09884-4>, and Yamaguchi and Martinez (2023) <doi:10.1111/bmsp.12308>. To facilitate their applications, 'variationalDCM' is developed to provide a collection of recently-proposed variational Bayesian estimation methods for various DCMs.