fit_rcbmm: Model selection

In ben-j-barlow/rcbmm: Regularized Copula-Based Mixture Model: Estimation and Model Selection

Description

The function selects the most appropriate model from a family of regularized copula-based mixture models arising from a varying number of components and a differing shirnkage parameter.

Usage

fit_rcbmm(
  x,
  lambda_grid,
  K = seq.int(2, 9),
  margins,
  maxit = 1000,
  epsilon = 1e-06,
  transform = TRUE,
  trace = FALSE
)

Arguments

x	A numeric matrix or data frame of observations. Rows correspond to observations and columns correspond to variables.
lambda_grid	An integer vector specifying the the values of the shrinkage parameter for which a regularized copula-based mixture model should be fitted. The default is lambda_grid = 0. If the vector parsed does not contain 0, then 0 is appended as starting values can only be obtained in the case no regularization is applied to the model.
K	An integer vector specifying the number of components for which a regularized copula-based mixture model should be fitted. The default is K=2:9.
margins	A character vector specifying the marginal distributions of the components in the mixture. The vector must have a length equal to the number of columns in x. Each element must be equal to "norm", "beta" or "gamma".
maxit	A numeric value specifying the maximum number of iterations the ECM algorithm should run before being halted.
epsilon	A numeric value indicting the tolerance for convergence.
transform	A logical value indicating whether or not starting values for the case lambda = 0 should be obtained using the transformations SPH, PCS, PCR and SVD. The default is TRUE.
trace	A logical value indicating if an update regarding the step's progress should be displayed.

Value

• BICA matrix demonstrating the BIC values achieved by each model in the family. Each column corresponds to a given value of lambda and each row corresponds to a given number of components.

• SILA matrix demonstrating the average silhouette width achieved by each model in the family. Each column corresponds to a given value of lambda and each row corresponds to a given number of components.

• all_modelsA list of lists with each element containing information about a specific model fitted.

• selected_modelsA list containing the information about the optimal model for each number of mixture components in the family. The optimal model for each number of componets is selected by picking the lambda that results in the largest average silhouette width. See the help file for silhouette for details.

• final_modelThe model contained in selected_models that maximized BIC.

See Also

ecm initialize.ecm

ben-j-barlow/rcbmm documentation built on Feb. 12, 2021, 9:14 a.m.