find_peaks {fossilbrush}

find_peaks

Description

Function to scan, column-wise, a matrix of per-taxon observation density time series. This can be applied to either the histogram or the kernel density output of densify, but the latter is recommended. Peaks are detected as local maxima, then smoothed within a local window and tested to distinguish if they are noise or significant. Strict threshold is that the peak is greater than the mean + sd of the window

Usage

find_peaks(x, win = 5, verbose = TRUE)

Arguments

x

A matrix as outputed by densify

win

A positive integer specifying the window length on either side of a peak (i.e. win 5 will give a total window of 11 - -5 indices + peak index + 5 indices)

verbose

A logical determining if function progress should be reported

Value

A list of four, the first three positions containing lists of the peak indices for each taxon, under raw, mean + sd and mean detection regimes. The fourth item is a dataframe of counts of peaks per taxon, 1 row per taxon, 1 column per detection regime

Examples

# load dataset
data("brachios")
# subsample brachios to make for a short example runtime
set.seed(1)
brachios <- brachios[sample(1:nrow(brachios), 1000),]
# get density matrix
dens <- densify(brachios)
# run function, using kernel density matrix
pk <- find_peaks(dens$kdensity)