Add function to extract activity level for a subrange of the daily cycle

[MarcusRowcliffe]

Standard errors of the estimate require boostrap trends to be saved by fitact function in actmod objects (the matrix will be largish c. 4 MB, but that's fine).

[MarcusRowcliffe]

#Nascent code for a function (or sub-function of fitact)
#Function takes a fitted 24-hour activity model (fit) and a 2-element vector (range) giving the radian times between which to calculate the percentage of time spent active.
partial.act <- function(fit, range){
pd <- head(fit@pdf[,2], -1) #extract prob densities and matching times (minus final overlapping value)
tt <- head(fit@pdf[,1], -1)
if(range[1]>range[2]){ #if range overlaps midnight...
tt[tt<range[2]] <- tt[tt<range[2]]+2pi #increment lower times and second range value to make range increasing
range[2] <- range[2]+2pi
}
i <- tt>range[1] & tt<range[2] #indicator, are times within the range?
sum(pd[i])2pi / (512 * (max(pd)*diff(rng))) #return activity level; note fit has 512 pdf values, so the approx. area under the curve is sum(pd)2pi/512
}

library(activity)
data(BCItime)
fit <- fitact(subset(BCItime, species=="ocelot")$time2pi)
plot(fit)
#Calculate activity levels for day and night
partial.act(fit, c(6,18)*pi/12)
partial.act(fit, c(18,6)*pi/12)