edma_fit: Nonparametric EDMA fit

In psolymos/EDMAinR: Euclidean Distance Matrix Analysis in R

Nonparametric EDMA fit

Description

Estimate mean form and SigmaKstar matrix based on Lele (1991), Lele and Richtsmeier (1991) and Hu (2007).

Usage

edma_fit(x, B=0, ncores=getOption("Ncpus", 1L))

## generics
Meanform(object, ...)
SigmaKstar(object, ...)
get_fm(object, ...)

## methods
## S3 method for class 'edma_fit_np'
print(x, truncate=40, ...)
## S3 method for class 'edma_fit'
Meanform(object, ...)
## S3 method for class 'edma_fit'
SigmaKstar(object, ...)
## S3 method for class 'edma_fit'
get_fm(object, sort=FALSE, level=0.95, ...)
## S3 method for class 'edma_fit'
confint(object, parm, level=0.95, ...)
## S3 method for class 'edma_fit'
as.edma_data(x, ...)

## plot methods
plot_3d(x, ...)
## S3 method for class 'edma_fit'
plot(x, ...)
## S3 method for class 'edma_fit'
plot_2d(x, ...)
## S3 method for class 'edma_fit'
plot_3d(x, ...)
## S3 method for class 'edma_fit'
plot_ord(x, ...)
## S3 method for class 'edma_fit'
plot_clust(x, ...)

## distance manipulation
## S3 method for class 'edma_fit'
as.dist(m, diag=FALSE, upper=FALSE)
## S3 method for class 'dist'
stack(x, ...)

Arguments

x, object, m	an EDMA data object of class 'edma_data'.
B	nonnegative integer, the number of bootstrap replicates.
ncores	positive integer, the number of cores to use when bootstrapping. Use options(Ncpus = 2) to set it to 2 globally.
truncate	numeric, number of characters to print for the object title.
sort	logical, if stacked distances are to be sorted, see Examples.
level	numeric, between 0 and 1, alpha level for confidence interval.
parm	a specification of which parameters are to be given confidence intervals, either a vector of numbers or a vector of names. See confint.
diag, upper	logical, indicating whether the diagonal and the upper triangle of the distance matrix should be printed. See as.dist.
...	other arguments passed to methods. E.g. for plot_clust, the method describing the clustering agglomeration method to be used by the link{hclust} function (default is "ward.D2").

Details

The function estimates mean form and SigmaKstar matrix based on Lele (1991), Lele and Richtsmeier (1991) and Hu (2007).

Value

edma_fit returns and EDMA fit object of class 'edma_fit'. .edma_fit_np is the workhorse function behind edma_fit.

stack.dist takes any distance matrix of class 'dist' and turns that into a long form data frame with columns row and col indicating the row and column labels, dist giving the value in that cell. Only returns the values from the lower triangle of the matrix.

get_fm is the intended user interface to extract the form matrix (FM) from EDMA fit objects. This has the stacked distances based on the mean form. When the object has bootstrap replicates, get_fm also returns confidence intervals for the distances based on bootstrap and the confint method.

Meanform extracts the mean form (K x D) matrix, SigmaKstar extracts the corresponding uncertainties (K x K) based on the EDMA fit object.

plot and plot_2d produces a 2D plot of the mean form. 2D projection is used in case of 3D landmark data based on metric multidimensional scaling. plot_3d use the rgl package to make a 3D plot. The sizes of the dots correspond to square root of the SigmaKstar diagonal elements.

The plot_ord and plot_clust produce plots based on dissimilarities among specimens, see plot_ord.edma_data for details.

Author(s)

Peter Solymos, Subhash R. Lele, Theodore M. Cole, Liangyuan Hu, Joan T. Richtsmeier

References

Lele, S. R., 1991. Some comments on coordinate-free and scale-invariant methods in morphometrics. American Journal of Physical Anthropology 85:407–417. <doi:10.1002/ajpa.1330850405>

Lele, S. R., and Richtsmeier, J. T., 1991. Euclidean distance matrix analysis: A coordinate-free approach for comparing biological shapes using landmark data. American Journal of Physical Anthropology 86(3):415–27. <doi:10.1002/ajpa.1330860307>

Hu, L., 2007. Euclidean Distance Matrix Analysis of Landmarks Data: Estimation of Variance. Thesis, Master of Science in Statistics, Department of Mathematical and Statistical Sciences, University of Alberta, Edmonton, Alberta, Canada. Pp. 49.

See Also

EDMA data: read_xyz

Form difference: edma_fdm

Growth difference: edma_gdm

Shape difference: edma_sdm

Examples

file <- system.file(
    "extdata/crouzon/Crouzon_P0_Global_MUT.xyz",
    package="EDMAinR")
x <- read_xyz(file)
x <- x[,,1:10] # 10 specimens

## nonparametric fit
fit <- edma_fit(x, B=9)
fit
str(Meanform(fit))
str(SigmaKstar(fit))

## form matrix
str(as.dist(fit))
str(stack(as.dist(fit)))

head(get_fm(fit))
head(get_fm(fit, sort=TRUE, decreasing=TRUE))
head(get_fm(fit, sort=TRUE, decreasing=FALSE))

plot_ord(fit)
plot_clust(fit)
plot(fit)
plot_2d(fit)
if (interactive())
    plot_3d(fit)

psolymos/EDMAinR documentation built on Aug. 25, 2023, 7:54 p.m.