bout-misc: Fit a Broken Stick Model on Log Frequency Data for...

In diveMove: Dive analysis and calibration

Description

Application of methods described by Sibly et al. (1990) and Mori et al. (2001) for the identification of bouts of behaviour.

Usage

boutfreqs(x, bw, method=c("standard", "seq.diff"), plot=TRUE, ...)
boutinit(lnfreq, x.break, plot=TRUE, ...)
labelBouts(x, bec, bec.method=c("standard", "seq.diff"))
logit(p)
unLogit(logit)

Arguments

x	numeric vector on which bouts will be identified based on “method”. For labelBouts it can also be a matrix with different variables for which bouts should be identified.
bw	numeric scalar: bin width for the histogram.
method, bec.method	character: method used for calculating the frequencies: “standard” simply uses x, while “seq.diff” uses the sequential differences method.
plot	logical, whether to plot results or not.
...	For boutfreqs, arguments passed to hist (must exclude breaks and include.lowest); for boutinit, arguments passed to plot (must exclude type).
lnfreq	data.frame with components lnfreq (log frequencies) and corresponding x (mid points of histogram bins).
x.break	vector of length 1 or 2 with x value(s) defining the break(s) point(s) for broken stick model, such that x < x.break[1] is 1st process, and x >= x.break[1] & x < x.break[2] is 2nd one, and x >= x.break[2] is 3rd one.
bec	numeric vector or matrix with values for the bout ending criterion which should be compared against the values in x for identifying the bouts.
p	numeric vector of proportions (0-1) to transform to the logit scale.
logit	numeric scalar: logit value to transform back to original scale.

Details

This follows the procedure described in Mori et al. (2001), which is based on Sibly et al. 1990. Currently, only a two process model is supported.

boutfreqs creates a histogram with the log transformed frequencies of x with a chosen bin width and upper limit. Bins following empty ones have their frequencies averaged over the number of previous empty bins plus one.

boutinit fits a "broken stick" model to the log frequencies modelled as a function of x (well, the midpoints of the binned data), using chosen value(s) to separate the two or three processes.

labelBouts labels each element (or row, if a matrix) of x with a sequential number, identifying which bout the reading belongs to. The bec argument needs to have the same dimensions as x to allow for situations where bec within x.

logit and unLogit are useful for reparameterizing the negative maximum likelihood function, if using Langton et al. (1995).

Value

boutfreqs returns a data frame with components lnfreq containing the log frequencies and x, containing the corresponding mid points of the histogram. Empty bins are excluded. A plot (histogram of input data) is produced as a side effect if argument plot is TRUE. See the Details section.

boutinit returns a list with as many elements as the number of processes implied by x.break (i.e. length(x.break) + 1). Each element is a vector of length two, corresponding to a and lambda, which are starting values derived from broken stick model. A plot is produced as a side effect if argument plot is TRUE.

labelBouts returns a numeric vector sequentially labelling each row or element of x, which associates it with a particular bout.

unLogit and logit return a numeric vector with the (un)transformed arguments.

Author(s)

Sebastian P. Luque [e-mail]

References

Langton, S.; Collett, D. and Sibly, R. (1995) Splitting behaviour into bouts; a maximum likelihood approach. Behaviour 132, 9-10.

Luque, S.P. and Guinet, C. (2007) A maximum likelihood approach for identifying dive bouts improves accuracy, precision, and objectivity. Behaviour, 144, 1315-1332.

Mori, Y.; Yoda, K. and Sato, K. (2001) Defining dive bouts using a sequential differences analysis. Behaviour, 2001 138, 1451-1466.

Sibly, R.; Nott, H. and Fletcher, D. (1990) Splitting behaviour into bouts. Animal Behaviour 39, 63-69.

See Also

bouts2.nls, bouts.mle. These include an example for labelBouts.

Examples

## Using the Example from '?diveStats':
utils::example("diveStats", package="diveMove",
               ask=FALSE, echo=FALSE)
postdives <- tdrX.tab$postdive.dur[tdrX.tab$phase.no == 2]
## Remove isolated dives
postdives <- postdives[postdives < 2000]
lnfreq <- boutfreqs(postdives, bw=0.1, method="seq.diff", plot=FALSE)
boutinit(lnfreq, 50)

## See ?bouts.mle for labelBouts() example

diveMove documentation built on May 2, 2019, 4:47 p.m.