Model-Based Clustering of Several Dissimilarity Matrices.

Description

The dmbc package implements a Bayesian algorithm for clustering a set of dissimilarity matrices within a model-based framework. In particular, we consider the case where S matrices are available, each describing the dissimilarities among n objects, possibly expressed by S subjects (judges), or measured under different experimental conditions, or with reference to different characteristics of the objects them- selves. Specifically, we focus on binary dissimilarities, taking values 0 or 1 depending on whether or not two objects are deemed as similar, with the goal of analyzing such data using multidimensional scaling (MDS). Differently from the standard MDS algorithms, we are interested in partitioning the dissimilarity matrices into clusters and, simultaneously, to extract a specific MDS configuration for each cluster. The parameter estimates are derived using a hybrid Metropolis-Gibbs Markov Chain Monte Carlo algorithm. We also include a BIC-like criterion for jointly selecting the optimal number of clusters and latent space dimensions.

For efficiency reasons, the core computations in the package are implemented using the C programming language and the RcppArmadillo package.

The dmbc package also supports the simulation of multiple chains through the support of the parallel package.

Plotting functionalities are imported from the nice bayesplot package. Currently, the package includes methods for binary data only. In future releases routines will be added specifically for continuous (i.e. normal), multinomial and count data.

dmbc classes

The dmbc package defines the following new classes:

dmbc_data:

defines the data to use in a DMBC model.

dmbc_model:

defines a DMBC model.

dmbc_fit:

defines the results of a DMBC analysis for a single MCMC chain.

dmbc_fit_list:

defines the results of a DMBC analysis for multiple MCMC chains.

dmbc_ic:

defines the results of the computation of the information criterion for a DMBC analysis.

dmbc_config:

defines the estimate of the latent configuration for a DMBC analysis.

The package includes print, summary and plot methods for each one of these classes.

Resources

Bug reports:

If you have noticed a bug that needs to be fixed, please let us know at the dmbc issue tracker on GitHub:

https://github.com/sergioventurini/dmbc/issues/.

General questions and help:

To ask a question about dmbc send and email to:

sergio.venturini@unicatt.it.

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

theme_default for the default ggplot theme used by bayesplot.

bayesplot-colors to set or view the color scheme used for plotting with bayesplot.

ggsave in ggplot2 for saving plots.

Adjustment of the center and orientation of a latent configuration.

Description

adjust_x adjusts the center and orientation of a latent configuration in Bayesian (metric) multidimensional scaling (BMDS).

Usage

adjust_x(x)

Arguments

Value

A list with elements:

x

A real matrix containing the adjusted latent configuration.

Sig_x

The variance and covariance matrix of the adjusted latent configuration.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

See Also

bmds for (one-way) Bayesian (metric) multidimensional scaling.

Examples

n <- 100
nr <- 20
nc <- floor(n/nr)
x <- matrix(rnorm(1:n), nrow = nr, ncol = nc)
adj_x <- adjust_x(x)
adj_x$x
adj_x$Sig_x

List of binary dissimilarity matrices among 18 animals.

Description

To illustrate the MDS analysis of sorting data, Takane et al. (2009) refer to judgments on the similarity between n = 18 animals expressed by S = 20 subjects. Each subject was asked to divide the animals into as many groups as needed, based on their similarity. We converted these values to 0 or 1 depending on whether a pair of animals is placed or not in the same group by a subject.

Usage

data(animals)

Format

A dmbc_data object whose diss element is a list of 20 binary dissimilarity matrices. Each matrix is defined as a dist object measuring whether each pair of the 18 animals has is placed in the same group (1) or not (0).

The dist objects have rows and columns that are named as follows:

be

bear

cm

camel

ct

cat

cw

cow

dg

dog

el

elephant

gf

giraffe

fx

fox

hs

horse

li

lion

mk

monkey

ms

mouse

pg

pig

rb

rabbit

sh

sheep

sq

squirrel

tg

tiger

wf

wolf

References

Takane, Y., Jung, S., Takane, Y. O. (2009). "Multidimensional Scaling". In Millsap, R. E., Maydeu-Olivares, A. (eds.), The SAGE Handbook of Quantitative Methods in Psychology, chapter 10, pp. 217–242,.

Examples

data(animals)
library(bayesplot)
cols <- color_scheme_set("teal")
plot(animals, colors = unlist(cols)[c(1, 6)], font = 1, cex.font = 0.75)

Bayesian multidimensional scaling (BMDS) using Markov Chain Monte Carlo (MCMC).

Description

bmds computes the Bayesian multidimensional scaling (BMDS) solutions using Markov Chain Monte Carlo for a range of specified latent space dimensions.

Usage

bmds(
  D,
  min_p = 1,
  max_pm1 = 6,
  burnin = 0,
  nsim = 13000,
  ic = TRUE,
  verbose = TRUE
)

Arguments

Value

A list with the following elements:

x.chain

MCMC chain of the latent configuration coordinates.

sigma.chain

MCMC chain of the random error.

lambda.chain

MCMC chain of the latent configuration variances.

stress

Numeric vector of the stress function values.

mdsIC

List with two elements, the MDSIC and BIC values for the required solutions.

accept

Numeric matrix of acceptance rates.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Oh, M.-S., Raftery, A. E. (2001), "Bayesian Multidimensional Scaling and Choice of Dimension", Journal of the American Statistical Association, 96, 1031-1044.

See Also

cmdscale for classical (metric) multidimensional scaling.

Examples

## Not run: 
# Airline Distances Between Cities
airline <- read.csv(file = system.file("extdata", "airline.csv",
  package = "dmbc"))
airline.nm <- airline[, 1]
airline <- airline[, 2:31]
colnames(airline) <- airline.nm
airline <- as.dist(airline)

min_p <- 1
max_p <- 4
burnin <- 200
nsim <- 1000
totiter <- burnin + nsim

airline.mds <- cmdscale(airline, max_p)
airline.bmds <- bmds(airline, min_p, max_p, burnin, nsim)

opar <- par(mfrow = c(1, 2))
plot(min_p:max_p, airline.bmds$mdsIC$mdsic, type = "b",
  main = "MDS Information Criterion", xlab = "p", ylab = "MDSIC")
MDSICmin <- which.min(airline.bmds$mdsIC$mdsic)
points((min_p:max_p)[MDSICmin], airline.bmds$mdsIC$mdsic[MDSICmin],
  col = "red", pch = 10, cex = 1.75, lwd = 1.5)

airline.bmds.x.mode <- bmds_get_x_mode(airline, airline.bmds, MDSICmin,
  min_p, max_p, start = (burnin + 1), end = totiter)
airline.bmds.d <- dist(airline.bmds.x.mode)
airline.mds.d <- dist(airline.mds[, 1:((min_p:max_p)[MDSICmin])])
plot(airline, airline.bmds.d, type = "n", xlab = "observed",
  ylab = "estimated", main = "Airline Distances \n Between Cities",
  xlim = c(0, max(airline, airline.bmds.d)),
  ylim = c(0, max(airline, airline.bmds.d)))
abline(0, 1, lty = 2, col = "gray")
points(airline, airline.mds.d, pch = 19, col = "cyan", cex = .5)
points(airline, airline.bmds.d, pch = 19, col = "magenta", cex = .5)
legend(x = "bottomright", legend = c("Classical MDS", "Bayesian MDS"),
  pch = c(19, 19), col = c("cyan", "magenta"))
par(opar)

# Careers of Lloyds Bank Employees
lloyds <- read.csv(file = system.file("extdata", "lloyds.csv",
  package = "dmbc"))
lloyds.nm <- lloyds[, 1]
lloyds <- lloyds[, 2:81]
colnames(lloyds) <- lloyds.nm
lloyds <- as.dist(lloyds)

min_p <- 1
max_p <- 12
burnin <- 200
nsim <- 1000
totiter <- burnin + nsim

lloyds.mds <- cmdscale(lloyds, max_p)
lloyds.bmds <- bmds(lloyds, min_p, max_p, burnin, nsim)

opar <- par(mfrow = c(1, 2))
plot((min_p:max_p), lloyds.bmds$mdsIC$mdsic, type = "b",
  main = "MDS Information Criterion", xlab = "p", ylab = "MDSIC")
MDSICmin <- which.min(lloyds.bmds$mdsIC$mdsic)
points((min_p:max_p)[MDSICmin], lloyds.bmds$mdsIC$mdsic[MDSICmin],
  col = "red", pch = 10, cex = 1.75, lwd = 1.5)

lloyds.bmds.x.mode <- bmds_get_x_mode(lloyds, lloyds.bmds, MDSICmin,
  min_p, max_p, start = (burnin + 1), end = totiter)
lloyds.bmds.d <- dist(lloyds.bmds.x.mode)
lloyds.mds.d <- dist(lloyds.mds[, 1:((min_p:max_p)[MDSICmin])])
plot(lloyds, lloyds.bmds.d, type = "n", xlab = "observed",
  ylab = "estimated", main = "Careers of Lloyds \n Bank Employees, 1905-1950",
  xlim = c(0, max(lloyds, lloyds.bmds.d)),
  ylim = c(0, max(lloyds, lloyds.bmds.d)))
abline(0, 1, lty = 2, col = "gray")
points(lloyds, lloyds.mds.d, pch = 19, col = "cyan", cex = .5)
points(lloyds, lloyds.bmds.d, pch = 19, col = "magenta", cex = .5)
legend(x = "topleft", legend = c("Classical MDS", "Bayesian MDS"),
  pch = c(19, 19), col = c("cyan", "magenta"))
par(opar)

# Road distances (in km) between 21 cities in Europe
data(eurodist, package = "datasets")

min_p <- 1
max_p <- 10
burnin <- 200
nsim <- 1000
totiter <- burnin + nsim

eurodist.mds <- cmdscale(eurodist, max_p)
eurodist.bmds <- bmds(eurodist, min_p, max_p, burnin, nsim)

opar <- par(mfrow = c(1, 2))
plot((min_p:max_p), eurodist.bmds$mdsIC$mdsic, type = "b",
  main = "MDS Information Criterion", xlab = "p", ylab = "MDSIC")
MDSICmin <- which.min(eurodist.bmds$mdsIC$mdsic)
points((min_p:max_p)[MDSICmin], eurodist.bmds$mdsIC$mdsic[MDSICmin],
  col = "red", pch = 10, cex = 1.75, lwd = 1.5)

eurodist.bmds.x.mode <- bmds_get_x_mode(eurodist, eurodist.bmds,
  MDSICmin, min_p, max_p, start = (burnin + 1), end = totiter)
eurodist.bmds.d <- dist(eurodist.bmds.x.mode)
eurodist.mds.d <- dist(eurodist.mds[, 1:((min_p:max_p)[MDSICmin])])
plot(eurodist, eurodist.bmds.d, type = "n", xlab = "observed",
  ylab = "estimated", main = "Road distances (in km) \n between 21 cities in Europe",
  xlim = c(0, max(eurodist, eurodist.bmds.d)),
  ylim = c(0, max(eurodist, eurodist.bmds.d)))
abline(0, 1, lty = 2, col = "gray")
points(eurodist, eurodist.mds.d, pch = 19, col = "cyan", cex = .5)
points(eurodist, eurodist.bmds.d, pch = 19, col = "magenta", cex = .5)
legend(x = "topleft", legend = c("Classical MDS", "Bayesian MDS"),
  pch = c(19, 19), col = c("cyan", "magenta"))
par(opar)

## End(Not run)

Posterior mode latent configuration in Bayesian multidimensional scaling (BMDS).

Description

bmds_get_x_mode returns the latent configuration that produced the largest posterior value during the MCMC.

Usage

bmds_get_x_mode(D, res, p.i, min_p, max_p, start, end)

Arguments

Value

A real matrix containing the posterior mode latent configuration.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

See Also

bmds for (one-way) Bayesian (metric) multidimensional scaling.

Examples

## Not run: 
# Airline Distances Between Cities
airline <- read.csv(file = system.file("extdata", "airline.csv",
  package = "dmbc"))
airline.nm <- airline[, 1]
airline <- airline[, 2:31]
colnames(airline) <- airline.nm
airline <- as.dist(airline)

min_p <- 1
max_p <- 4
burnin <- 200
nsim <- 1000
totiter <- burnin + nsim

airline.mds <- cmdscale(airline, max_p)
airline.bmds <- bmds(airline, min_p, max_p, burnin, nsim)

opar <- par(mfrow = c(1, 2))
plot(min_p:max_p, airline.bmds$mdsIC$mdsic, type = "b",
  main = "MDS Information Criterion", xlab = "p", ylab = "MDSIC")
MDSICmin <- which.min(airline.bmds$mdsIC$mdsic)
points((min_p:max_p)[MDSICmin], airline.bmds$mdsIC$mdsic[MDSICmin],
  col = "red", pch = 10, cex = 1.75, lwd = 1.5)

airline.bmds.x.mode <- bmds_get_x_mode(airline, airline.bmds, MDSICmin,
  min_p, max_p, start = (burnin + 1), end = totiter)
airline.bmds.d <- dist(airline.bmds.x.mode)
airline.mds.d <- dist(airline.mds[, 1:((min_p:max_p)[MDSICmin])])
plot(airline, airline.bmds.d, type = "n", xlab = "observed",
  ylab = "estimated", main = "Airline Distances \n Between Cities",
  xlim = c(0, max(airline, airline.bmds.d)),
  ylim = c(0, max(airline, airline.bmds.d)))
abline(0, 1, lty = 2, col = "gray")
points(airline, airline.mds.d, pch = 19, col = "cyan", cex = .5)
points(airline, airline.bmds.d, pch = 19, col = "magenta", cex = .5)
legend(x = "bottomright", legend = c("Classical MDS", "Bayesian MDS"),
  pch = c(19, 19), col = c("cyan", "magenta"))
par(opar)

## End(Not run)

Auxiliary function to recursively check NAs in a list.

Description

check_list_na() compares two lists and fills in the missing elements in the first with those included in the second. The comparison is recursive in the sense that the process is repeated for all lists included in those given.

Usage

check_list_na(orig, des)

Arguments

Value

A list with all elements added.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Examples

G <- 5
prior <- list(eta = list(a = rep(1, G), b = rep(2, G)))
check_list_na(prior, dmbc_prior())

Extract the final cluster memberships from a dmbc_config class instance.

Description

Extract the final cluster memberships from a dmbc_config class instance.

Usage

## S4 method for signature 'dmbc_config'
clusters(object, newdata = NULL, ...)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Sum of squared residuals (SSR) from the observed distances and the given latent configuration.

Description

comp_ssr computes the sum of squared residuals (SSR) from the observed distances (diss) and the given latent coordinates (x).

Usage

comp_ssr(x, diss)

Arguments

Value

A length-one numeric vector providing the SSR for its arguments.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

See Also

bmds for (one-way) Bayesian (metric) multidimensional scaling.

Examples

n <- 10000
nr <- 200
nc <- floor(n/nr)
x <- matrix(rnorm(1:n), nrow = nr, ncol = nc)
obsdiss <- dist(x)
ssr <- numeric(ncol(x))
for (i in 1:ncol(x)) {
  ssr[i] <- comp_ssr(x[, 1:i], obsdiss)
}
plot(ssr, xlab = "number of dimensions", ylab = "SSR", type = "b")

Estimation of a DMBC model.

Description

dmbc(), the main function of the package, estimates a DMBC model for a given set of S dissimilarity matrices.

Usage

dmbc(
  data,
  p = 2,
  G = 3,
  control = dmbc_control(),
  prior = NULL,
  cl = NULL,
  post_all = FALSE
)

Arguments

Value

A dmbc_fit_list object.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

bmds for Bayesian (metric) multidimensional scaling.

dmbc_data for a description of the data format.

dmbc_fit_list for a description of the elements included in the returned object.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 3
p <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 20000
nsim <- 10000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], random.start = TRUE, verbose = TRUE,
  nchains = 2, thin = 10, store.burnin = TRUE, threads = 2,
  parallel = "snow")
sim.dmbc <- dmbc(simdiss, p, G, control)

summary(sim.dmbc, include.burnin = FALSE)

library(bayesplot)
library(ggplot2)
color_scheme_set("teal")
plot(sim.dmbc, what = "trace", regex_pars = "eta")

z <- dmbc_get_configuration(sim.dmbc, chain = 1, est = "mean",
  labels = 1:16)
summary(z)
color_scheme_set("mix-pink-blue")
graph <- plot(z, size = 2, size_lbl = 3)
graph + panel_bg(fill = "gray90", color = NA)

## End(Not run)

Model selection of DMBC models.

Description

dmbc_IC() is the main function for simultaneously selecting the optimal latent space dimension (p) and number of clusters (G) for a DMBC analysis.

Usage

dmbc_IC(
  data,
  pmax = 3,
  Gmax = 5,
  control = dmbc_control(),
  prior = NULL,
  est = "mean"
)

Arguments

Value

A dmbc_ic object.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc() for fitting a DMBC model.

dmbc_ic for a description of the elements included in the returned object.

Examples

## Not run: 
data(simdiss, package = "dmbc")

pmax <- 2
Gmax <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 2000
nsim <- 1000
seed <- 1809

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], random.start = TRUE, verbose = TRUE,
  thin = 10, store.burnin = TRUE)
sim.ic <- dmbc_IC(data = simdiss, pmax = pmax, Gmax = Gmax, control = control,
  est = "mean")

pmax <- pmax + 1
Gmax <- Gmax + 2
new.ic <- update(sim.ic, pmax = pmax, Gmax = Gmax)
new.ic

# plot the results
library(bayesplot)
library(ggplot2)
color_scheme_set("mix-yellow-blue")
p <- plot(new.ic, size = c(4, 1.5))
p + panel_bg(fill = "gray90", color = NA)

## End(Not run)

Auxiliary function for checking the grouping results of a fitted DMBC model.

Description

dmbc_check_groups() is an auxiliary function for checking whether the cluster membership estimates provided by the individual chains of the fitted model provided agree or not.

Usage

dmbc_check_groups(res, est = "mean")

Arguments

Value

A length-one logical vector, which is equal to TRUE if all simulated chains provide the same cluster membership estimates, and FALSE otherwise.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc_get_configuration() for a description of the configuration extractor function.

dmbc_fit_list for a description of a fitted DMBC model.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 3
p <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 2000
nsim <- 1000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], random.start = TRUE, verbose = TRUE,
  nchains = 2, thin = 10, store.burnin = TRUE, threads = 2,
  parallel = "snow")
sim.dmbc <- dmbc(simdiss, p, G, control)

dmbc_check_groups(sim.dmbc)

## End(Not run)

An S4 class to represent the latent configuration estimate for a DMBC model.

Description

An S4 class to represent the the latent configuration estimate for a DMBC model.

Slots

Z.est

An array containing the estimate of the latent configuration for a DMBC model.

Z.sd

An array containing the standard deviation of the latent configuration for a DMBC model.

cluster

A numeric vector providing the estimated group membership for the S subjects in the data.

est

A length-one character vector providing the estimate type returned in Z.est. Possible values are mean (posterior mean), median (posterior median), ml (maximum likelihood) and map (maximum-a-posteriori).

n

A length-one numeric vector providing the number of objects.

p

A length-one numeric vector providing the number of latent dimensions.

S

A length-one numeric vector providing the number of subjects.

G

A length-one numeric vector providing the number of clusters.

family

An object of class list; named list with elements representing the parameter estimates corresponding to different values of p and G.

chain

A length-one numeric vector representing the ID of the MCMC chain used to compute the estimates.

labels

A character vector for the (optional) strings to use in the plots for labeling the objects.

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

Examples

showClass("dmbc_config")

Auxiliary Function for Controlling DMBC Model Fitting

Description

dmbc_control() is an auxiliary function as user interface for dmbc() fitting. Typically only used when calling the dmbc() function. It is used to set parameters that affect the sampling but do not affect the posterior distribution.

control_dmbc() is an alias for dmbc_control().

check_control() is an auxiliary function that verifies the correctness of the controls provided before a DMBC is fitted with dmbc().

Usage

dmbc_control(
  nsim = 5000,
  burnin = 10000,
  thin = 1,
  nchains = 1,
  threads = 1,
  seed = NULL,
  parallel = "no",
  z.prop = 1.5,
  alpha.prop = 0.75,
  random.start = TRUE,
  partition = NULL,
  method = "manhattan",
  procrustes = TRUE,
  relabel = TRUE,
  store.burnin = TRUE,
  verbose = FALSE
)

control_dmbc(
  nsim = 5000,
  burnin = 10000,
  thin = 1,
  nchains = 1,
  threads = 1,
  seed = NULL,
  parallel = "no",
  z.prop = 1.5,
  alpha.prop = 0.75,
  random.start = TRUE,
  partition = NULL,
  method = "manhattan",
  procrustes = TRUE,
  relabel = TRUE,
  store.burnin = TRUE,
  verbose = FALSE
)

check_control(control)

Arguments

Value

A named list with the control options as components.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

See Also

dmbc()

Examples

## Not run: 
data(simdiss, package = "dmbc")
# Shorter run than default.
sim.fit <- dmbc(simdiss,
  control = dmbc_control(burnin = 1000, nsim = 2000, thin = 5, verbose = TRUE))

## End(Not run)

An S4 class to represent the data to use in a DMBC model.

Description

An S4 class to represent the data to use in a DMBC model.

Slots

diss

A list whose elements are the dissimilarity matrices corresponding to the judgments expressed by the S subjects/raters. These matrices must be defined as a dist object.

n

A length-one character vector representing the number of objects compared by each subject.

S

A length-one numeric vector representing the number of subjects.

family

A length-one character vector representing the type of data to analyze. Currently, it accepts only the 'binomial' value, but future developments will include the possibility to analyze continuous, multinomial and count data.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

Examples

showClass("dmbc_data")

Fitter function for DMBC models.

Description

dmbc_fit() is the main function that estimates a DMBC model.

Usage

dmbc_fit(D, p, G, family, control, prior, start)

Arguments

Value

A dmbc_fit_list object.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc_data for a description of the data format.

dmbc_fit_list for a description of the elements included in the returned object.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 3
p <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 20000
nsim <- 10000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], random.start = TRUE, verbose = TRUE,
  nchains = 2, thin = 10, store.burnin = TRUE, threads = 2,
  parallel = "snow")
sim.dmbc <- dmbc(simdiss, p, G, control)

summary(sim.dmbc, include.burnin = FALSE)

library(bayesplot)
library(ggplot2)
color_scheme_set("teal")
plot(sim.dmbc, what = "trace", regex_pars = "eta")

z <- dmbc_get_configuration(sim.dmbc, chain = 1, est = "mean",
  labels = 1:16)
summary(z)
color_scheme_set("mix-pink-blue")
graph <- plot(z, size = 2, size_lbl = 3)
graph + panel_bg(fill = "gray90", color = NA)

## End(Not run)

An S4 class to represent the results of fitting DMBC model.

Description

An S4 class to represent the results of fitting DMBC model using a single Markov Chain Monte Carlo chain.

Slots

z.chain

An object of class array; posterior draws from the MCMC algorithm for the (untransformed) latent configuration Z.

z.chain.p

An object of class array; posterior draws from the MCMC algorithm for the (Procrustes-transformed) latent configuration Z.

alpha.chain

An object of class matrix; posterior draws from the MCMC algorithm for the \alpha parameters.

eta.chain

An object of class matrix; posterior draws from the MCMC algorithm for the \eta parameters.

sigma2.chain

An object of class matrix; posterior draws from the MCMC algorithm for the \sigma^2 parameters.

lambda.chain

An object of class matrix; posterior draws from the MCMC algorithm for the \lambda parameters.

prob.chain

An object of class array; posterior draws from the MCMC algorithm for the cluster membership probabilities.

x.ind.chain

An object of class array; posterior draws from the MCMC algorithm for the cluster membership indicators.

x.chain

An object of class matrix; posterior draws from the MCMC algorithm for the cluster membership labels.

accept

An object of class matrix; final acceptance rates for the MCMC algorithm.

diss

An object of class list; list of observed dissimilarity matrices.

dens

An object of class list; list of log-likelihood, log-prior and log-posterior values at each iteration of the MCMC simulation.

control

An object of class list; list of the control parameters (number of burnin and sample iterations, number of MCMC chains, etc.). See dmbc_control() for more information.

prior

An object of class list; list of the prior hyperparameters. See dmbc_prior() for more information.

dim

An object of class list; list of dimensions for the estimated model, i.e. number of objects (n), number of latent dimensions (p), number of clusters (G), and number of subjects (S).

model

An object of class dmbc_model.

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

Examples

showClass("dmbc_fit")

An S4 class to represent the results of fitting DMBC model.

Description

An S4 class to represent the results of fitting DMBC model using multiple Markov Chain Monte Carlo chains.

Slots

results

An object of class list; list of dmbc_fit objects corresponding to the parallel MCMC chains simulated during the estimation.

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc_fit for more details on the components of each element of the list.

Examples

showClass("dmbc_fit_list")

Conversion of an dmbc_fit_list object to a list.

Description

dmbc_fit_list_to_list converts an object of class dmbc_fit_list to a list of arrays including all the parameter. chains. It is intended for internal use mainly.

Usage

dmbc_fit_list_to_list(res, include.burnin = FALSE, verbose = TRUE)

Arguments

Value

An object of type mcmc.list.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

See Also

dmbc() for for fitting a DMBC model; dmbc_fit_list-class.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 3
p <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 2000
nsim <- 1000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], nchains = 2, verbose = TRUE)
sim.dmbc <- dmbc(simdiss, p, G, control)
sim.list <- dmbc_fit_list_to_list(sim.dmbc, TRUE)

library(bayesplot)
mcmc_trace(sim.list, regex_pars = "lambda")

## End(Not run)

Conversion of an dmbc_fit_list object to an object of class mcmc.list.

Description

dmbc_fit_list_to_mcmc.list converts an object of class dmbc_fit_list to an object with class mcmc.list.

Usage

dmbc_fit_list_to_mcmc.list(res, include.burnin = FALSE, verbose = TRUE)

Arguments

Value

An object of type mcmc.list.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

See Also

dmbc() for for fitting a DMBC model; dmbc_fit_list-class; mcmc.list.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 3
p <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 2000
nsim <- 1000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], nchains = 2, verbose = TRUE)
sim.dmbc <- dmbc(simdiss, p, G, control)
sim.mcmc <- dmbc_fit_list_to_mcmc.list(sim.dmbc, TRUE)
plot(sim.mcmc)

## End(Not run)

Conversion of an dmbc_fit object to an object of class mcmc.

Description

dmbc_fit_to_mcmc converts an object of class dmbc_fit to an object with class mcmc.

Usage

dmbc_fit_to_mcmc(res, include.burnin = FALSE, verbose = TRUE)

Arguments

Value

An object of type mcmc.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

See Also

dmbc() for for fitting a DMBC model; dmbc_fit-class; mcmc.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 3
p <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 2000
nsim <- 1000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], verbose = TRUE)
sim.dmbc <- dmbc(simdiss, p, G, control)
sim.mcmc <- dmbc_fit_to_mcmc(sim.dmbc@results[[1]], TRUE)
plot(sim.mcmc)

## End(Not run)

Extractor function for a fitted DMBC model.

Description

dmbc_get_configuration() is an extractor function for extracting the latent configuration estimates of a fitted DMBC model.

Usage

dmbc_get_configuration(res, chain = 1, est = "mean", labels = character(0))

Arguments

Value

A dmbc_config object.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc_data for a description of the data format.

dmbc_fit_list for a description of the elements included in the returned object.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 3
p <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 2000
nsim <- 1000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], random.start = TRUE, verbose = TRUE,
  nchains = 2, thin = 10, store.burnin = TRUE, threads = 2,
  parallel = "snow")
sim.dmbc <- dmbc(simdiss, p, G, control)

z <- dmbc_get_configuration(sim.dmbc, chain = 1, est = "mean")
summary(z)

library(bayesplot)
library(ggplot2)
color_scheme_set("mix-pink-blue")
graph <- plot(z, size = 2, size_lbl = 3)
graph + panel_bg(fill = "gray90", color = NA)

## End(Not run)

Extractor function for a fitted DMBC model.

Description

dmbc_get_map() is an extractor function for extracting the maximum-a-posterior estimates of the parameters for a fitted DMBC model.

Usage

dmbc_get_map(res, chain = 1)

Arguments

Value

A named list with the following elements:

z:

array of latent coordinates posterior mean estimates

alpha:

numeric vector of alpha posterior mean estimates

eta:

numeric vector of eta posterior mean estimates

sigma2:

numeric vector of sigma2 posterior mean estimates

lambda:

numeric vector of lambda posterior mean estimates

prob:

numeric matrix of probability posterior mean estimates

cluster:

numeric vector of cluster membership posterior mean estimates

logpost:

length-one numeric vector of the maximum log-posterior value

chain:

length-one numeric vector of the MCMC chain number used

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc_data for a description of the data format.

dmbc_fit_list for a description of the elements included in the returned object.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 3
p <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 2000
nsim <- 1000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], random.start = TRUE, verbose = TRUE,
  nchains = 2, thin = 10, store.burnin = TRUE, threads = 2,
  parallel = "snow")
sim.dmbc <- dmbc(simdiss, p, G, control)

dmbc_get_map(sim.dmbc, chain = 1)

## End(Not run)

Extractor function for a fitted DMBC model.

Description

dmbc_get_ml() is an extractor function for extracting the maximum likelihood estimates of the parameters for a fitted DMBC model.

Usage

dmbc_get_ml(res, chain = 1)

Arguments

Value

A named list with the following elements:

z:

array of latent coordinates posterior mean estimates

alpha:

numeric vector of alpha posterior mean estimates

eta:

numeric vector of eta posterior mean estimates

sigma2:

numeric vector of sigma2 posterior mean estimates

lambda:

numeric vector of lambda posterior mean estimates

prob:

numeric matrix of probability posterior mean estimates

cluster:

numeric vector of cluster membership posterior mean estimates

loglik:

length-one numeric vector of the maximum log-likelihood value

chain:

length-one numeric vector of the MCMC chain number used

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc_data for a description of the data format.

dmbc_fit_list for a description of the elements included in the returned object.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 3
p <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 2000
nsim <- 1000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], random.start = TRUE, verbose = TRUE,
  nchains = 2, thin = 10, store.burnin = TRUE, threads = 2,
  parallel = "snow")
sim.dmbc <- dmbc(simdiss, p, G, control)

dmbc_get_ml(sim.dmbc, chain = 1)

## End(Not run)

Extractor function for a fitted DMBC model.

Description

dmbc_get_postmean() is an extractor function for extracting the posterior mean estimates of the parameters for a fitted DMBC model.

Usage

dmbc_get_postmean(res, chain = 1)

Arguments

Value

A named list with the following elements:

z:

array of latent coordinates posterior mean estimates

alpha:

numeric vector of alpha posterior mean estimates

eta:

numeric vector of eta posterior mean estimates

sigma2:

numeric vector of sigma2 posterior mean estimates

lambda:

numeric vector of lambda posterior mean estimates

prob:

numeric matrix of probability posterior mean estimates

cluster:

numeric vector of cluster membership posterior mean estimates

chain:

length-one numeric vector of the MCMC chain number used

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc_data for a description of the data format.

dmbc_fit_list for a description of the elements included in the returned object.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 3
p <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 2000
nsim <- 1000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], random.start = TRUE, verbose = TRUE,
  nchains = 2, thin = 10, store.burnin = TRUE, threads = 2,
  parallel = "snow")
sim.dmbc <- dmbc(simdiss, p, G, control)

dmbc_get_postmean(sim.dmbc, chain = 1)

## End(Not run)

Extractor function for a fitted DMBC model.

Description

dmbc_get_postmedian() is an extractor function for extracting the posterior median estimates of the parameters for a fitted DMBC model.

Usage

dmbc_get_postmedian(res, chain = 1)

Arguments

Value

A named list with the following elements:

z:

array of latent coordinates posterior median estimates

alpha:

numeric vector of alpha posterior median estimates

eta:

numeric vector of eta posterior median estimates

sigma2:

numeric vector of sigma2 posterior median estimates

lambda:

numeric vector of lambda posterior median estimates

prob:

numeric matrix of probability posterior median estimates

cluster:

numeric vector of cluster membership posterior median estimates

chain:

length-one numeric vector of the MCMC chain number used

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc_data for a description of the data format.

dmbc_fit_list for a description of the elements included in the returned object.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 3
p <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 2000
nsim <- 1000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], random.start = TRUE, verbose = TRUE,
  nchains = 2, thin = 10, store.burnin = TRUE, threads = 2,
  parallel = "snow")
sim.dmbc <- dmbc(simdiss, p, G, control)

dmbc_get_postmedian(sim.dmbc, chain = 1)

## End(Not run)

An S4 class to represent the comparison of a set of DMBC models.

Description

An S4 class to represent the comparison of a set of DMBC models through the dissimilarity model-based clustering information criterion (DCIC).

Slots

logprior

An object of class matrix providing the log-prior values corresponding to different values of p and G.

logmlik

An object of class matrix providing the marginal log-likelihood values corresponding to different values of p and G.

logcorrfact

An object of class matrix providing the logarithm of the correction factors corresponding to different values of p and G.

DCIC

An object of class matrix providing the values of the DCIC index corresponding to different values of p and G.

post.est

An object of class list; named list with elements representing the parameter estimates corresponding to different values of p and G.

est

A length-one character vector representing the estimate type used in computing the DCIC index. Possible values are mean, median, ml and map. See dmbc_ic() for more details about these values.

res_last_p

An object of class list; list of dmbc_fit_list objects with the results of fitting the DMBC models corresponding to the last value of p. This is needed in case of an update of the DCIC calculations using additional p and/or G values.

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

Examples

showClass("dmbc_ic")

Function to compute the starting values before fitting a DMBC models.

Description

dmbc_init() is the main function that estimates a DMBC model.

Usage

dmbc_init(D, p, G, family, random.start, method, partition)

Arguments

Value

A named list with the following items:

z:

array of latent coordinates starting values

x:

numeric vector of initial cluster memberships

ng:

numeric vector of initial cluster sizes

alpha:

numeric vector of alpha starting values

eta:

numeric vector of eta starting values

sigma2:

numeric vector of sigma2 starting values

lambda:

numeric vector of lambda starting values

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc() for fitting a DMBC model.

Examples

data(simdiss, package = "dmbc")
dmbc_init(simdiss@diss, p = 2, G = 3, family = "binomial", random.start = TRUE)

Log-likelihood for DMBC models.

Description

dmbc_logLik() computes the log-likelihood value for a DMBC model.

Usage

dmbc_logLik(D, Z, alpha, lambda, x)

Arguments

Value

A length-one numeric vector of the log-likelihood value.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc().

Log-likelihood for DMBC models.

Description

dmbc_logLik_rbmds() computes the log-likelihood value for a DMBC model.

Usage

dmbc_logLik_rbmds(D, Z, alpha)

Arguments

Value

A length-one numeric vector of the log-likelihood value.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc().

Auxiliary function for realigning the grouping of a fitted DMBC model.

Description

dmbc_match_groups() is an auxiliary function for realigning the cluster membership estimates provided by the individual chains of the fitted model if they do not agree.

Usage

dmbc_match_groups(res, est = "mean", ref = 1)

Arguments

Value

An object of class dmbc_fit_list.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc_check_groups() for checking the consistency of the cluster memberships across chains for a fitted DMBC model.

dmbc_get_configuration() for a description of the configuration extractor function.

dmbc_fit_list for a description of a fitted DMBC model.

Examples

## Not run: 
data(simdiss, package = "dmbc")

G <- 5
p <- 3
prm.prop <- list(z = 4, alpha = 2)
burnin <- 2000
nsim <- 1000
seed <- 2301

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], random.start = TRUE, verbose = TRUE,
  nchains = 6, store.burnin = TRUE, threads = 2, parallel = "snow")
sim.dmbc <- dmbc(simdiss, p, G, control)

sim.dmbc_new <- dmbc_match_groups(sim.dmbc)

## End(Not run)

An S4 class to represent a DMBC model.

Description

An S4 class to represent a DMBC model.

Slots

p

A length-one character vector representing the number of dimensions of the latent space to use in the MDS analysis.

G

A length-one numeric vector representing the number of clusters to partition the subjects into.

family

A length-one character vector representing the type of data to analyze. Currently, it accepts only the 'binomial' value, but future developments will include the possibility to analyze continuous, multinomial and count data.

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

Examples

showClass("dmbc_model")

Auxiliary Function for Setting DMBC Model Priors

Description

dmbc_prior() is an auxiliary function as user interface for dmbc() fitting. Typically only used when calling the dmbc() function. It is used to set prior hyperparameters.

prior_dmbc() is an alias for dmbc_prior().

check_prior() is an auxiliary function that verifies the correctness of the prior hyperparameters provided before a DMBC is fitted with dmbc().

update_prior() is an auxiliary function to modify a set of prior choices using a new value of p and G. It is intended for internal use mainly in the dmbc_ic() function.

Usage

dmbc_prior(
  eta = list(a = rep(1.5, .dmbcEnv$current_G), b = rep(0.5, .dmbcEnv$current_G)),
  sigma2 = list(a = 0.1, b = 0.1),
  lambda = rep(1, .dmbcEnv$current_G)
)

prior_dmbc(
  eta = list(a = rep(1.5, .dmbcEnv$current_G), b = rep(0.5, .dmbcEnv$current_G)),
  sigma2 = list(a = 0.1, b = 0.1),
  lambda = rep(1, .dmbcEnv$current_G)
)

check_prior(prior)

update_prior(prior, p, G)

Arguments

Value

A list with the prior hyperparameters as components.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

See Also

dmbc()

Examples

## Not run: 
data(simdiss, package = "dmbc")
# Shorter run than default.
sim.fit <- dmbc(simdiss,
  control = dmbc_control(burnin = 1000, nsim = 2000, thin = 1, verbose = TRUE),
  prior = dmbc_prior(sigma2 = list(a = 1, b = 4)))

## End(Not run)

Create an instance of the dmbc_config class using new/initialize.

Description

Create an instance of the dmbc_config class using new/initialize.

Usage

## S4 method for signature 'dmbc_config'
initialize(
  .Object,
  Z.est = array(),
  Z.sd = array(),
  cluster = numeric(),
  est = character(),
  n = numeric(),
  S = numeric(),
  p = numeric(),
  G = numeric(),
  family = character(),
  chain = numeric(),
  labels = character()
)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Create an instance of the dmbc_data class using new/initialize.

Description

Create an instance of the dmbc_data class using new/initialize.

Usage

## S4 method for signature 'dmbc_data'
initialize(
  .Object,
  diss = list(),
  n = numeric(),
  S = numeric(),
  family = character()
)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Create an instance of the dmbc_fit class using new/initialize.

Description

Create an instance of the dmbc_fit class using new/initialize.

Usage

## S4 method for signature 'dmbc_fit'
initialize(
  .Object,
  z.chain = array(),
  z.chain.p = array(),
  alpha.chain = matrix(),
  eta.chain = matrix(),
  sigma2.chain = matrix(),
  lambda.chain = matrix(),
  prob.chain = array(),
  x.ind.chain = array(),
  x.chain = matrix(),
  accept = matrix(),
  diss = list(),
  dens = list(),
  control = list(),
  prior = list(),
  dim = list(),
  model = NA
)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Create an instance of the dmbc_fit_list class using new/initialize.

Description

Create an instance of the dmbc_fit_list class using new/initialize.

Usage

## S4 method for signature 'dmbc_fit_list'
initialize(.Object, results = list())

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Create an instance of the dmbc_ic class using new/initialize.

Description

Create an instance of the dmbc_ic class using new/initialize.

Usage

## S4 method for signature 'dmbc_ic'
initialize(
  .Object,
  logprior = matrix(),
  logmlik = matrix(),
  logcorrfact = matrix(),
  DCIC = matrix(),
  post.est = list(),
  est = character(),
  res_last_p = list()
)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Create an instance of the dmbc_model class using new/initialize.

Description

Create an instance of the dmbc_model class using new/initialize.

Usage

## S4 method for signature 'dmbc_model'
initialize(.Object, p = numeric(), G = numeric(), family = character())

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

List of binary dissimilarity matrices among 15 kinship terms.

Description

Rosenberg and Kim (1975) designed an experiment to analyze the perceived similarities of 15 kinship terms.

Here, we consider the data relative to 85 females made available in Rosenberg (1982). Each subject was asked to group the kinship terms according to the perceived similarity. Thus, S = 85 binary dissimilarity matrices are available whose elements (0 or 1) indicate whether or not two kinship terms were grouped together by each individual.

Usage

data(kinship)

Format

A dmbc_data object whose diss element is a list of 85 binary dissimilarity matrices. Each matrix is defined as a dist object measuring whether each pair of the 15 kinship terms is judged as similar (1) or not (0).

The dist objects have rows and columns that are named as follows:

GrF

grandfather

GrM

grandmother

GrD

granddaughter

GrS

grandson

Bro

brother

Sis

sister

Fat

father

Mot

mother

Dau

daughter

Son

son

Nep

nephew

Nie

niece

Cou

cousin

Aun

aunt

Unc

uncle

References

Rosenberg, S. (1982). The method of sorting in multivariate research with applications selected from cognitive psychology and person perception. In N Hirschberg, LG Humphreys (eds.), Multivariate Applications in the Social Sciences, pp. 117–142. Erlbaum., Hillsdale, NJ.

Rosenberg, S., Kim, M. P. (1975). The method of sorting as a data-gathering procedure in multivariate research. Multivariate Behavioral Research, 10.

Examples

data(kinship)
library(bayesplot)
cols <- color_scheme_set("mix-red-blue")
plot(kinship, colors = unlist(cols)[c(1, 6)], font = 1, cex.font = 0.75)

Information criterion for Bayesian multidimensional scaling (BMDS).

Description

mdsic computes the information criterion for a set of Bayesian multidimensional scaling (BMDS) solutions using the approach in Oh & Raftery (2001).

Usage

mdsic(x_star, rmin_ssr, n, min_p = 1, max_p = 6)

Arguments

Value

A list with the following elements:

mdsic

A numeric vector with the values of MDSIC index.

bic

A numeric vector with the values of the BIC index.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Oh, M.-S., Raftery, A. E. (2001), "Bayesian Multidimensional Scaling and Choice of Dimension", Journal of the American Statistical Association, 96, 1031-1044.

See Also

bmds for Bayesian (metric) multidimensional scaling and comp_ssr for the computation of SSR.

Examples

## Not run: 
# Road distances (in km) between 21 cities in Europe
data(eurodist, package = "datasets")

min_p <- 1
max_p <- 10
burnin <- 200
nsim <- 1000
totiter <- burnin + nsim

eurodist.mds <- cmdscale(eurodist, max_p)
eurodist.bmds <- bmds(eurodist, min_p, max_p, burnin, nsim)

plot((min_p:max_p), eurodist.bmds$mdsIC$mdsic, type = "b",
  main = "MDS Information Criterion", xlab = "p", ylab = "MDSIC")
MDSICmin <- which.min(eurodist.bmds$mdsIC$mdsic)
points((min_p:max_p)[MDSICmin], eurodist.bmds$mdsIC$mdsic[MDSICmin],
  col = "red", pch = 10, cex = 1.75, lwd = 1.5)

## End(Not run)

Provide a graphical summary of a dmbc_config class instance.

Description

Provide a graphical summary of a dmbc_config class instance.

Usage

## S4 method for signature 'dmbc_config,ANY'
plot(
  x,
  size = NULL,
  size_lbl = NULL,
  nudge_x = 0,
  nudge_y = 0,
  label_objects = TRUE,
  ...
)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Provide a graphical summary of a dmbc_data class instance.

Description

Provide a graphical summary of a dmbc_data class instance.

Usage

## S4 method for signature 'dmbc_data,ANY'
plot(x, colors = c("white", "black"), font = NA, cex.font = NA, ...)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Examples

data(simdiss)
library(bayesplot)
cols <- color_scheme_set("brightblue")
plot(simdiss, colors = unlist(cols)[c(1, 6)], font = 1, cex.font = 0.75)

Provide a graphical summary of a dmbc_fit class instance.

Description

Provide a graphical summary of a dmbc_fit class instance.

Usage

## S4 method for signature 'dmbc_fit,ANY'
plot(
  x,
  what = "trace",
  pars = character(),
  regex_pars = "lambda",
  include.burnin = FALSE,
  combo = NULL,
  ...
)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Provide a graphical summary of a dmbc_fit_list class instance.

Description

Provide a graphical summary of a dmbc_fit_list class instance.

Usage

## S4 method for signature 'dmbc_fit_list,ANY'
plot(
  x,
  what = "trace",
  pars = character(),
  regex_pars = "lambda",
  include.burnin = FALSE,
  combo = NULL,
  ...
)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Provide a graphical summary of a dmbc_ic class instance.

Description

Provide a graphical summary of a dmbc_ic class instance.

Usage

## S4 method for signature 'dmbc_ic,ANY'
plot(x, size = NULL, ...)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Show an instance of the dmbc_config class.

Description

Show an instance of the dmbc_config class.

Usage

## S4 method for signature 'dmbc_config'
show(object)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Show an instance of the dmbc_data class.

Description

Show an instance of the dmbc_data class.

Usage

## S4 method for signature 'dmbc_data'
show(object)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Show an instance of the dmbc_fit class.

Description

Show an instance of the dmbc_fit class.

Usage

## S4 method for signature 'dmbc_fit'
show(object)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Show an instance of the dmbc_fit_list class.

Description

Show an instance of the dmbc_fit_list class.

Usage

## S4 method for signature 'dmbc_fit_list'
show(object)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Show an instance of the dmbc_ic class.

Description

Show an instance of the dmbc_ic class.

Usage

## S4 method for signature 'dmbc_ic'
show(object)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Show an instance of the dmbc_model class.

Description

Show an instance of the dmbc_model class.

Usage

## S4 method for signature 'dmbc_model'
show(object)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Simulated binary dissimilarity matrices.

Description

A dataset containing a list of simulated binary dissimilarity matrices.

Usage

data(simdiss)

Format

A dmbc_data object whose diss element is a list of 10 binary dissimilarity matrices. Each matrix is defined as a dist object measuring the agreement among 16 different units.

Examples

data(simdiss)
library(bayesplot)
cols <- color_scheme_set("brightblue")
plot(simdiss, colors = unlist(cols)[c(1, 6)], font = 1, cex.font = 0.75)

Subsetting a dmbc_fit object.

Description

Subsetting a dmbc_fit object.

Usage

## S4 method for signature 'dmbc_fit'
subset(x, pars = character(), regex_pars = character(), ...)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Subsetting a dmbc_fit_list object.

Description

Subsetting a dmbc_fit_list object.

Usage

## S4 method for signature 'dmbc_fit_list'
subset(x, pars = character(), regex_pars = character(), ...)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Provide a summary of a dmbc_config class instance.

Description

Provide a summary of a dmbc_config class instance.

Usage

## S4 method for signature 'dmbc_config'
summary(object)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Provide a summary of a dmbc_data class instance.

Description

Provide a summary of a dmbc_data class instance.

Usage

## S4 method for signature 'dmbc_data'
summary(object)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Provide a summary of a dmbc_fit class instance.

Description

Provide a summary of a dmbc_fit class instance.

Usage

## S4 method for signature 'dmbc_fit'
summary(object, include.burnin = FALSE, summary.Z = FALSE, ...)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Provide a summary of a dmbc_fit_list class instance.

Description

Provide a summary of a dmbc_fit_list class instance.

Usage

## S4 method for signature 'dmbc_fit_list'
summary(object, include.burnin = FALSE, summary.Z = FALSE, ...)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Provide a summary of a dmbc_ic class instance.

Description

Provide a summary of a dmbc_ic class instance.

Usage

## S4 method for signature 'dmbc_ic'
summary(object, p = NULL, G = NULL)

Arguments

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

Provide an update of a dmbc_ic class instance.

Description

Provide an update of a dmbc_ic class instance.

Usage

## S4 method for signature 'dmbc_ic'
update(object, pmax = NULL, Gmax = NULL, ...)

Arguments

Value

A dmbc_ic object.

Author(s)

Sergio Venturini sergio.venturini@unicatt.it

References

Venturini, S., Piccarreta, R. (2021), "A Bayesian Approach for Model-Based Clustering of Several Binary Dissimilarity Matrices: the dmbc Package in R", Journal of Statistical Software, 100, 16, 1–35, <10.18637/jss.v100.i16>.

See Also

dmbc() for fitting a DMBC model.

dmbc_ic for a description of the elements included in the returned object.

Examples

## Not run: 
data(simdiss, package = "dmbc")

pmax <- 2
Gmax <- 2
prm.prop <- list(z = 1.5, alpha = .75)
burnin <- 2000
nsim <- 1000
seed <- 1809

set.seed(seed)

control <- list(burnin = burnin, nsim = nsim, z.prop = prm.prop[["z"]],
  alpha.prop = prm.prop[["alpha"]], random.start = TRUE, verbose = TRUE,
  thin = 10, store.burnin = TRUE)
sim.ic <- dmbc_IC(data = simdiss, pmax = pmax, Gmax = Gmax, control = control,
  est = "mean")

pmax <- pmax + 1
Gmax <- Gmax + 2
new.ic <- update(sim.ic, pmax = pmax, Gmax = Gmax)
new.ic

# plot the results
library(bayesplot)
library(ggplot2)
color_scheme_set("mix-yellow-blue")
p <- plot(new.ic, size = c(4, 1.5))
p + panel_bg(fill = "gray90", color = NA)

## End(Not run)