R/fitNMDS.R

In phytomosaic/fitNMDS: Fitting Resampled and Reciprocal NMDS Models

#' @description Measures of agreement for bootstrapped and reciprocal
#'     NMDS.
#' @details Combining two different datasets into one nonmetric
#'     multidimensional scaling (NMDS) model can be risky if they each
#'     cover different attribute spaces (e.g., different species pools
#'     in ecology). Therefore, comparing two datasets requires
#'     estimating internal agreement (sampling variability) relative
#'     to external agreement (exchangeability).
#'
#'     Resampled NMDS with \code{resamp_nmds} estimates internal
#'     agreement (sampling variability) of one candidate dataset.
#'     The premise of this is to resample dissimilarities (either
#'     through jackknife or bootstrap resampling), perform NMDS
#'     ordination on each resampled replicate, then determine
#'     collective agreement across all NMDS solutions.  Other uses of
#'     resampled NMDS may include estimating confidence regions for
#'     ordination site scores, or testing the stability of species
#'     positions along ordination axes.
#'
#'     Reciprocal NMDS with \code{recip_nmds} estimates external
#'     agreement (exchangeability) of two candidate datasets.  The
#'     premise of reciprocal NMDS is to calibrate each of two models
#'     with each respective dataset, mutually exchanging datasets
#'     among the alternative ordination models, then determining
#'     measures of how well each dataset fit the alternative model.
#'     This is really just a special case of 2-fold cross-validation
#'     as applied to NMDS.  The primary use of reciprocal NMDS is to
#'     test the null hypothesis that two multivariate datasets are
#'     exchangeable.
#'
#' @docType package
#' @name fitNMDS-package
#' @title The fitNMDS Package for Evaluating Nonmetric
#'     Multidimensional Scaling (NMDS) Models
#' @keywords package
NULL