cvTune: Tuning Parameter Selection via Cross-Validation

In dilernia/rccm: Implemention of Random Covariance Models

Tuning Parameter Selection via Cross-Validation

Description

This function selects the optimal set of tuning parameters for the Random Covariance Clustering Model (RCCM) based on cross-validation.

Usage

cvTune(
  x,
  G,
  lambs,
  methods = c("RCCM", "GGL", "GLasso", "FGL", "RCM"),
  folds = 5
)

Arguments

x	List of K data matrices each of dimension n_k x p
G	Number of groups or clusters.
lambs	A data frame of candidate tuning parameter values with three columns: lambda1, lambda2, and lambda3.
methods	Methods to implement cross-validation for. Must be a vector containing one or more of "FGL", "GGL", "GLasso", "RCCM", or "RCM".
folds	Number of folds to use for cross-validation.

Value

A data frame of optimally selected tuning parameter values and the correspond average negative log-likelihoods across the folds number of folds with five columns: lambda1, lambda2, lambda3, nll, and method.

Author(s)

Andrew DiLernia

Examples

# Generate data with 2 clusters with 12 and 10 subjects respectively,
# 15 variables for each subject, 100 observations for each variable for each subject,
# the groups sharing about 50% of network connections, and 10% of differential connections
# within each group
set.seed(1994)
G <- 2
myData <- rccSim(G = G, clustSize = 10, p = 10, n = 177, overlap = 0.20, rho = 0.10)

# Analyze simulated data with RCCM & GLasso using 5-fold CV
lambdas <- expand.grid(lambda1 = c(5, 10, 15), lambda2 = c(50, 100), lambda3 = 0.10)
cvLambdas <- cvTune(x = myData$simDat, G = G, lambs = lambdas, methods = c("RCCM", "GLasso"), folds = 5)

# RCCM with CV selected tuning parameters
resRCCM <- rccm(x = myData$simDat, lambda1 = cvLambdas[which(cvLambdas$method == "RCCM"), ]$lambda1,
                lambda2 = cvLambdas[which(cvLambdas$method == "RCCM"), ]$lambda2,
                lambda3 = cvLambdas[which(cvLambdas$method == "RCCM"), ]$lambda3, nclusts = 2)

# GLasso with CV selected tuning parameters
resGLasso <- lapply(myData$simDat, FUN = function(x) {
glasso::glasso(cov(x), rho = cvLambdas[which(cvLambdas$method == "GLasso"), ]$lambda1,
penalize.diagonal = FALSE)$wi})

dilernia/rccm documentation built on Sept. 25, 2022, 9:40 a.m.