LCPA - A General Framework for Latent Classify and Profile Analysis
A unified latent class modeling framework that encompasses
both latent class analysis (LCA) and latent profile analysis
(LPA), offering a one-stop solution for latent class modeling.
It implements state-of-the-art parameter estimation methods,
including the expectation–maximization (EM) algorithm, neural
network estimation (NNE; requires users to have 'Python' and
its dependent libraries installed on their computer), and
integration with 'Mplus' (requires users to have 'Mplus'
installed on their computer). In addition, it provides commonly
used model fit indices such as the Akaike information criterion
(AIC) and Bayesian information criterion (BIC), as well as
classification accuracy measures such as entropy. The package
also includes fully functional likelihood ratio tests (LRT) and
bootstrap likelihood ratio tests (BLRT) to facilitate model
comparison, along with bootstrap-based and observed information
matrix-based standard error estimation. Furthermore, it
supports the standard three-step approach for LCA, LPA, and
latent transition analysis (LTA) with covariates, enabling
detailed covariate analysis. Finally, it includes several
user-friendly auxiliary functions to enhance interactive
usability.
