unmarkedMultFrame.RdThese functions construct unmarkedFrames for data collected during primary and secondary sampling periods.
unmarkedMultFrame(y, siteCovs, obsCovs, numPrimary, yearlySiteCovs)
unmarkedFrameGMM(y, siteCovs, obsCovs, numPrimary, yearlySiteCovs, type,
obsToY, piFun)
unmarkedFrameGDS(y, siteCovs, numPrimary, yearlySiteCovs, dist.breaks,
survey, unitsIn, tlength)
unmarkedFrameGPC(y, siteCovs, obsCovs, numPrimary, yearlySiteCovs)unmarkedMultFrame objects are used by colext.
unmarkedFrameGMM objects are used by gmultmix.
unmarkedFrameGDS objects are used by gdistsamp.
unmarkedFrameGPC objects are used by gpcount.
For a study with M sites, T years, and a maximum of
J observations per site-year, the data can be supplied in a
variety of ways but are stored as follows.
y is an \(M \times TJ\) matrix, with each row
corresponding to a site. siteCovs is a data frame with \(M\)
rows. yearlySiteCovs is a data frame with \(MT\) rows which
are in site-major, year-minor order. obsCovs is a data frame
with \(MTJ\) rows, which are ordered by site-year-observation, so that
a column of obsCovs corresponds to as.vector(t(y)),
element-by-element. The number of years must be specified in
numPrimary.
If the data are in long format, the convenience function
formatMult is useful for creating the unmarkedMultFrame.
unmarkedFrameGMM and unmarkedFrameGDS are superclasses of
unmarkedMultFrame containing information on
the survey design used that resulted in multinomial outcomes. For
unmarkedFrameGMM and constant-interval removal sampling, you can set
type="removal" and ignore
the arguments obsToY and piFun. Similarly, for double-observer sampling,
setting type="double" or type="depDouble" will automatically create an appropiate
obsToY matrix and piFuns. For all other situations, the type
argument of unmarkedFrameGMM should be
ignored and the obsToY and piFun arguments must be specified. piFun must be a
function that converts an MxJ matrix of detection probabilities into an MxJ
matrix of multinomial cell probabilities. obsToY is a matrix describing how
the obsCovs relate to the observed counts y. For further discussion and examples
see the help page for multinomPois and piFuns.
unmarkedFrameGMM and unmarkedFrameGDS objects can be created from an unmarkedMultFrame using the "as" conversion method. See examples.
Data used with colext, gmultmix, and gdistsamp may be collected during a single year, so yearlySiteCovs may be a misnomer is some cases.
an unmarkedMultFrame or unmarkedFrameGMM object
A matrix of the observed data.
Data frame of covariates that vary at the site level.
Data frame of covariates that vary within site-year-observation level.
Number of primary time periods (seasons in the multiseason model).
Data frame containing covariates at the site-year level.
Set to "removal" for constant-interval removal sampling, "double" for standard double observer sampling, or "depDouble" for dependent double observer sampling. This should be not be specified for other types of survey designs.
A matrix specifying relationship between observation-level covariates and response matrix
A function converting an MxJ matrix of detection probabilities into an MxJ matrix of multinomial cell probabilities.
see unmarkedFrameDS
see unmarkedFrameDS
see unmarkedFrameDS
see unmarkedFrameDS
n <- 50 # number of sites
T <- 4 # number of primary periods
J <- 3 # number of secondary periods
site <- 1:50
years <- data.frame(matrix(rep(2010:2013, each=n), n, T))
years <- data.frame(lapply(years, as.factor))
occasions <- data.frame(matrix(rep(1:(J*T), each=n), n, J*T))
y <- matrix(0:1, n, J*T)
umf <- unmarkedMultFrame(y=y,
siteCovs = data.frame(site=site),
obsCovs=list(occasion=occasions),
yearlySiteCovs=list(year=years),
numPrimary=T)
#> Warning: yearlySiteCovs contains characters. Converting them to factors.
umfGMM1 <- unmarkedFrameGMM(y=y,
siteCovs = data.frame(site=site),
obsCovs=list(occasion=occasions),
yearlySiteCovs=data.frame(year=c(t(years))),
# or: yearlySiteCovs=list(year=years),
numPrimary=T, type="removal")
#> Warning: yearlySiteCovs contains characters. Converting them to factors.
# A user-defined piFun calculating removal probs when time intervals differ.
instRemPiFun <- function(p) {
M <- nrow(p)
J <- ncol(p)
pi <- matrix(NA, M, J)
p[,1] <- pi[,1] <- 1 - (1 - p[,1])^2
p[,2] <- 1 - (1 - p[,2])^3
p[,3] <- 1 - (1 - p[,3])^5
for(i in 2:J) {
pi[,i] <- pi[, i - 1]/p[, i - 1] * (1 - p[, i - 1]) * p[, i]
}
return(pi)
}
# Associated obsToY matrix required by unmarkedFrameMPois
o2y <- diag(ncol(y))
o2y[upper.tri(o2y)] <- 1
o2y
#> [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10] [,11] [,12]
#> [1,] 1 1 1 1 1 1 1 1 1 1 1 1
#> [2,] 0 1 1 1 1 1 1 1 1 1 1 1
#> [3,] 0 0 1 1 1 1 1 1 1 1 1 1
#> [4,] 0 0 0 1 1 1 1 1 1 1 1 1
#> [ reached getOption("max.print") -- omitted 8 rows ]
umfGMM2 <- unmarkedFrameGMM(y=y,
siteCovs = data.frame(site=site),
obsCovs=list(occasion=occasions),
yearlySiteCovs=data.frame(year=c(t(years))),
numPrimary=T, obsToY=o2y, piFun="instRemPiFun")
#> Warning: yearlySiteCovs contains characters. Converting them to factors.
str(umfGMM2)
#> Formal class 'unmarkedFrameGMM' [package "unmarked"] with 9 slots
#> ..@ piFun : chr "instRemPiFun"
#> ..@ samplingMethod: chr "userDefined"
#> ..@ numPrimary : num 4
#> ..@ yearlySiteCovs:'data.frame': 200 obs. of 1 variable:
#> .. ..$ year: Factor w/ 4 levels "2010","2011",..: 1 2 3 4 1 2 3 4 1 2 ...
#> ..@ y : int [1:50, 1:12] 0 1 0 1 0 1 0 1 0 1 ...
#> ..@ obsCovs :'data.frame': 600 obs. of 1 variable:
#> .. ..$ occasion: int [1:600] 1 2 3 4 5 6 7 8 9 10 ...
#> ..@ siteCovs :'data.frame': 50 obs. of 1 variable:
#> .. ..$ site: int [1:50] 1 2 3 4 5 6 7 8 9 10 ...
#> ..@ mapInfo : NULL
#> ..@ obsToY : num [1:12, 1:12] 1 0 0 0 0 0 0 0 0 0 ...