difference: Estimate the proximity of two individuals

In ctmm: Continuous-Time Movement Modeling

Estimate the proximity of two individuals

Description

Given a pair of telemetry objects and ctmm movement models, predict their location differences at shared times and estimate their mean-square distance.

Usage

difference(data,CTMM,t=NULL,...)

distances(data,CTMM,t=NULL,level=0.95,...)

proximity(data,CTMM,GUESS=ctmm(error=TRUE),debias=TRUE,level=0.95,...)

Arguments

data	A list of two telemetry objects.
CTMM	A list of two ctmm movement-model objects.
t	An optional vector of times over which to predict the location differences.
level	Confidence level for the distance/proximity estimate.
GUESS	An optional ctmm object to specify the candidate model parameters of the location differences.
debias	Include inverse-χ^2 bias corrections.
...	Options passed to ctmm.select.

Details

The difference function predicts the location difference vectors, (x_A-x_B,y_A-y_B), for a pair of individuals, \{A,B\}, at overlapping times. The distances function further estimates the instantaneous distances between individuals. The proximity function fits an autocorrelation model to the output of difference, and then compares the mean-square distance between the individuals to what you would expect if the two individuals were moving independently.

Value

difference outputs a telemetry object of the location differences with prediction covariances. distances outputs a data.frame of distance estimates with confidence intervals. proximity outputs a ratio estimate with confidence intervals, where values <1 indiciate that the two individuals are closer on average than expected for independent movement, 1 is consistent with independent movement, and values >1 indicate that the individuals are farther from each other on average than expected for independent movement. Therefore, if the CIs contain 1, then the distance is insignificant with a p-value threshold of 1-level (two-sided) or half that for a one-sided test.

Author(s)

C. H. Fleming.

See Also

ctmm.select, predict.ctmm

Examples

#Load package
library(ctmm)

# load buffalo data
data(buffalo)

# select two buffalo that overlap in space and time
DATA <- buffalo[c(1,3)]
# plot the two buffalo
plot(DATA,col=c('red','blue'))

FITS <- list()
for(i in 1:2)
{
  GUESS <- ctmm.guess(DATA[[i]],interactive=FALSE)
  # in general, you want to use ctmm.select
  FITS[[i]] <- ctmm.fit(DATA[[i]],GUESS)
}

# calculate difference vectors
DIFF <- difference(DATA,FITS)
# plot the difference vectors with prediction-error ellipses
plot(DIFF)

# calculate the proximity statistic
# disabling location error for speed
proximity(DATA,FITS,GUESS=ctmm(error=FALSE))

ctmm documentation built on Nov. 4, 2022, 5:06 p.m.