Package 'plotmm'

Tidy Visualization of a Cut Point from a Mixture Model

Description

Returns a plot of the data density (histogram) with an overlaid cut point generated by the fit mixture model

Usage

plot_cut_point(m, plot = TRUE, color = c("grayscale", "amerika", "wesanderson"))

Arguments

m	An object of class mixEM corresponding with the fit mixture model
plot	Logical for generating the plot. If FALSE, only the cut point value from the GMM is returned. If TRUE, histogram with the overlaid cut point is returned. Default is set to TRUE.
color	A vector of color options including "amerika" (from amerika package), "wesanderson" (from wesanderson package), and "grayscale", which is the default option.

Details

Mixture models can be used to derive cut points separating clusters via soft assignment (See Benaglia et al. 2009 for more). plot_cut_point() plots data density with an overlaid cut point (the mean of the calculated mu) from mixEM objects via mixtools. Note, this function is in its infancy, and at present only works in the limited context of 2-component Gaussian mixture models with equal priors and equal variances. Development for more complex cases is in process.

References

Benaglia, T., Chauveau, D., Hunter, D. and Young, D. 2009. mixtools: An R package for analyzing finite mixture models. Journal of Statistical Software, 32(6), pp.1-29.

Ram, K., and Wickham, H. 2015. wesanderson: a Wes Anderson palette generator. R package version 0.3.

Examples

if(require(mixtools)){
mixmdl <- mixtools::normalmixEM(faithful$waiting, k = 2)
}
plot_cut_point(mixmdl, plot = TRUE, color = "amerika") # returns plot, amerika
plot_cut_point(mixmdl, plot = TRUE, color = "wesanderson") # returns plot, wesanderson
plot_cut_point(mixmdl, plot = FALSE) # returns only the cut point value from the GMM

---

Plots Mixture Components from Gaussian Mixture Models

Description

Generates a plot of data densities with overlaid mixture components from a Gaussian mixture model (GMM)

Usage

plot_gmm(m, k = NULL)

Arguments

m	An object of class mixEM corresponding with the fit GMM
k	The number of components specified in the GMM, m

Details

Original function from the plotGMM package. Retained here for bridging between the packages. We recommend using instead the updated plot_mm function.

Note: plot_gmm requires a mixtools object to be supplied. Users must enter the same component value, k, in the plot_gmm function, as that which was specified in the original GMM specification (also k in mixtools).

References

Benaglia, T., Chauveau, D., Hunter, D. and Young, D., 2009. mixtools: An R package for analyzing finite mixture models. Journal of Statistical Software, 32(6), pp.1-29.

Wickham, H., 2016. ggplot2: elegant graphics for data analysis. Springer.

Examples

if(require(mixtools)){
mixmdl <- mixtools::normalmixEM(faithful$waiting, k = 2)
}
plot_gmm(mixmdl, 2)

---

Helper Function for Overlaying Mixture Components

Description

Allows for plotting mixture components conditioned on a superimposed function meant for passage to ggplot's stat_function()

Usage

plot_mix_comps(x, mu = NULL, sigma = NULL, lam = 1, beta0 = NULL,
 beta1=NULL, alpha=NULL, beta=NULL,
 normal=FALSE, logisreg=FALSE,
 gamma=FALSE, poisson=FALSE)

Arguments

x	Input data
mu	Component mean
sigma	Component variance
lam	Component mixture weight
beta0	Coefficient values
beta1	Coefficient values
alpha	Initial shape parameters
beta	Initial parameter values
normal	Logical for normal distribution
logisreg	Logical for logistic regression mixtures
gamma	Logical for gamma distribution
poisson	Logical for poisson regression mixtures

Details

Allows for component curves to be superimposed over a mixture model plot

Examples

if(require(mixtools)){
mixmdl <- mixtools::normalmixEM(faithful$waiting, k = 2)
}
x <- mixmdl$x
x <- data.frame(x)
ggplot2::ggplot(data.frame(x)) +
  ggplot2::geom_density(ggplot2::aes(x), color="black", fill="black") +
  ggplot2::stat_function(geom = "line", fun = plot_mix_comps,
                args = list(mixmdl$mu[1], mixmdl$sigma[1], lam = mixmdl$lambda[1]),
                colour = "red") +
  ggplot2::stat_function(geom = "line", fun = plot_mix_comps,
                args = list(mixmdl$mu[2], mixmdl$sigma[2], lam = mixmdl$lambda[2]),
               colour = "blue")

---

Custom Function for Overlaying Mixture Components for Normal Distributions

Description

Plots a mixture component conditioned on a superimposed function

Usage

plot_mix_comps_normal(x, mu, sigma, lam)

Arguments

x	Input data
mu	Mean of component
sigma	Variance of component
lam	Mixture weight of component

Details

Allows for specifying a custom function to be superimposed when plotting a mixture component assuming a normal distribution. This is the original function for the package, which is also included in the updated plot_mix_comps() function.

Examples

if(require(mixtools)){
mixmdl <- mixtools::normalmixEM(faithful$waiting, k = 2)
}
x <- mixmdl$x
x <- data.frame(x)
ggplot2::ggplot(data.frame(x)) +
  ggplot2::geom_density(ggplot2::aes(x), color="black", fill="black") +
  ggplot2::stat_function(geom = "line", fun = plot_mix_comps_normal,
                args = list(mixmdl$mu[1], mixmdl$sigma[1], lam = mixmdl$lambda[1]),
                colour = "red") +
  ggplot2::stat_function(geom = "line", fun = plot_mix_comps_normal,
                args = list(mixmdl$mu[2], mixmdl$sigma[2], lam = mixmdl$lambda[2]),
               colour = "blue")

---

Tidy Visualization of Mixture Models

Description

Generates a ggplot of data densities with overlaid mixture components from fit mixture models.

Usage

plot_mm(m, k = NULL, data = NULL)

Arguments

m	A mixture model object
k	Optional. The number of components specified in the mixture model, m
data	Name of data object required only for EMCluster objects

Details

This is the core function in the package, returning a ggplot object for a fit mixture model. The plot includes the data density with overlaid mixture components.

References

Wickham, H., 2016. ggplot2: elegant graphics for data analysis. Springer.

Examples

if(require(mixtools)){
mixmdl1 <- mixtools::normalmixEM(faithful$waiting, k = 2)
}
plot_mm(mixmdl1, 2)

if(require(mixtools)){
x <- c(rgamma(200, shape = 50, scale = 11), rgamma(200, shape = 28, scale = 6))
mixmdl2 <- mixtools::gammamixEM(x, lambda = c(1, 1)/2)
}
plot_mm(mixmdl2)

---