diagnostic-shc: plot diagnostics for shc object

In pkimes/sigclust2: sigclust2: Statistical Significance of Clustering

Description

Provides visualizations to check the null Gaussian assumption at a specified range of nodes along the dendrogram

Usage

## S3 method for class 'shc'
diagnostic(obj, K = 1, fname = NULL, ci_idx = 1,
  pty = "all", ...)

Arguments

obj	a shc object
K	an integer value or vector specifying the range of nodes for which to create diagnostic plots, with 1 corresponding to the root (default = 1)
fname	a character string specifying the name of the output file, see details for more information on default behavior depending on length(K) (default = NULL)
ci_idx	a numeric value between 1 and length(obj$in_args$ci) specifying which cluster index to use for creating plots (default = 1)
pty	a string specifying the plot type that should be created see details for more information on different plot types available (default = "all")
...	other parameters passed to plots

Details

If K is a single value, the default behavior is to output the diagnostic plot to the current device. This behavior is overriden if fname is specified by the user. If length(K) > 1, than the diagnostic plots are printed to paste0(fname, ".pdf"). If fname is not specified, "shc_diagnostic", is used by default.

The pty parameter accepts the following options:

• "background": empirical distribution of marginal data used for background variance estimation

• "qq": qqplot for checking null Gaussian assumption

• "covest": screeplot for checking covariance factor model

• "pvalue": empirical distribution of simulated CIs

• "all": all of the above plots (default).

Value

Prints plots to paste0(fname, ".pdf") if fname is specified or length(K)>1.

Author(s)

Patrick Kimes

pkimes/sigclust2 documentation built on May 25, 2019, 8:20 a.m.