mdshelper's Introduction

MDShelper

The goal of MDShelper is to provide some convenience functions when conducting multidimensional-scaling (MDS) analysis. So far the package contains the following functions (use ?function_name for more information):

cross_validation_MDS(): Runs a cross-validation procedure to find the best fitting number of dimensions of the MDS space.
BIC_MDS(): Computes the BIC for a specific number of dimensions of the MDS space according to Lee (2001)
gen_data_MDS(): Simulates data by generating a matrix with the true underlying dimensions and item coordinates, as well as the corresponding true pairwise distances
fill_mat(): Transforms a distance matrix with only lower triangle entries (rest NA) to full matrix with 0 on the diagonals
compute_dists(): Wrapper around a Rcpp-function (pair_dist_cpp) to compute pairwise distances between two vectors.

Installation

You can install the development version of MDShelper from GitHub with:

# install.packages("devtools")
devtools::install_github("dizyd/MDShelper")

Example

This is a basic example which shows you how generate pairwise distance data based on a defined number of underlying dimensions and running a cross-validation procedure to try to recover this number of dimensions

library(MDShelper)

## Generate pairwise distance matrix
sim_data <- gen_data_MDS(ndims = 4, n = 16)

## Run Cross-Validation
cross_validation_MDS(sim_data$dist_mat, max_dim = 5)

Recommend Projects