DecomposeR (version 1.0.2)

parsimony: Parsimony of a decomposition

Description

The function additions the absolute values of each component of a decomposition by depth/time, and computes the ratio of that with the absolute values of the signal. This is done either by depth/time or on the time/depth-cumulated signal (i.e. the bulk signal).

This is a proxy for parsimony: it is the factor of amplitude added by the decomposition. A perfect decomposition, that does not 'invent' wiggles, should approach 1, but will logically always be higher. However it is influenced by the absolute value of the initial signal: if the original signal is not centered around 0, the parsimony is not significative (it will artificially be closer to 1). To correct for that, the residue (part of the decomposition that is not centered around zero) has to be removed from the original signal.

Usage

parsimony(
  emd = NULL,
  xy = NULL,
  m = NULL,
  mode = NULL,
  repl = 1,
  bulk = TRUE,
  correct = NA
)

Arguments

emd

an emd object

xy

the signal

m

a matrix with columns of same length that xy, made of the decomposition of the signal

mode

the mode sequence index to give to each replicated IMFs

repl

the replication of decompositions in m

bulk

whether to have a bulk value each decomposition replication, or for each dt of each replication

correct

the modes to remove from the original signal and decomposition for a significative parsimony calculation. If NA, it removes the last mode, considered as the residue. Can be a vector of several integers, standing for the columns of m. If NULL, no mode is removed

Value

a matrix with each column being a replication, or a list of bulk values for each replication

Examples

# NOT RUN {
set.seed(42)

n <- 500

dt <- seq_len(n)
xy <- rnorm(n, mean = 0, sd = 1) + 10

dec <- extricate(xy, dt, nimf = 7, comb = 10,
                 factor_noise = 1, speak = TRUE)
# }
# NOT RUN {
plot_emd(dec, dir = tempdir())
# }
# NOT RUN {
parsimony(dec, correct = NULL)

parsimony(dec)

# }