classify_kmeans {musclesyneRgies}

Muscle synergy classification with k-means

Description

Muscle synergy classification with k-means

Usage

classify_kmeans(x, MSE_lim = 0.001, inspect = FALSE, show_plot = FALSE)

Arguments

x

A list of musclesyneRgies objects

MSE_lim

Mean squared error threshold for determining the minimum number of clusters

inspect

Logical, ask for interactive re-ordering or go fully automated?

show_plot

Logical, to decide whether plots should be plotted in the active graphic device

Details

This function must be applied to a list with a sufficient amount of trials, otherwise the classification will not work. Typically, at least 10 trials for the same condition are needed for satisfactory classification. If show_plot is TRUE (default) plots are also shown in the active graphic device. Plots can then be saved with the preferred export method, such as ggplot2::ggsave. The algorithm used is the default for stats::kmeans (Hartigan and Wong, 1979), which is known for its robustness to local minima. Nonetheless, the stochastic nature of the algorithm should prompt the user to attempt a few classifications and analyse their stability, before drawing conclusions on e.g. the number of fundamental synergies and/or their function. While the default parameters are optimised for human locomotion, it is suggested to test the function with different mean squared error thresholds, which is a crucial quantity to determine the number of clusters. Inspection and plotting are as well highly recommended to gain more insight into the classification process.

Value

List of musclesyneRgies objects, each with elements:

• syns factorisation rank or minimum number of synergies
  

• M motor modules (time-invariant coefficients)
  

• P motor primitives (time-dependent coefficients)
  

• V original data
  

• Vr reconstructed data
  

• iterations number of iterations to convergence
  

• R2 quality of reconstruction (coefficient of determination)

• rank_type was the rank fixed or variable?
  

• classification classification type (k-means)

Examples

# Load some data
data(SYNS)
# Classify synergies
SYNS_classified <- classify_kmeans(SYNS)