Description Usage Arguments Details Value Author(s) See Also Examples
Bayesian Model To Identify Factors Affecting Wildlife-Vehicle Collisions
1 | prepareFit(X, alphas, collisions, nYear, Area, departement)
|
X |
a data.frame containing the numeric variables supposed to have an effect on the wildlife-vehicle collisions (columns) for spatial unit (rows). |
alphas |
a character string vector with length equal to
|
collisions |
an integer vector with length equal to
|
nYear |
an integer vector with length equal to |
Area |
a numeric vector with length equal to |
departement |
a character vector with length equal to
|
prepareFit
prepares the elements required for the
fit of the Kuo-Mallik model (a Bayesian model used to identify
factors affecting wildlife-vehicle collisions).
a list with all elements required for the fit of the model
with JAGS, that is: (i) data4jags
: the list of the data
required by the model, to be passed to the argument data
of
the function jags.model
of the package rjags
, (ii)
ini
: list of starting values for the parameters, to be
passed to the argument init
of the function
jags.model
, (iii) modelstring
: a character string
containing the model fit by JAGS, (iv) coefnames
: vector of
character strings containing the names of the coefficients of
interest in the model, to be passed to the argument
variable.names
of the function coda.samples
of the
package rjags
.
Clement Calenge, clement.calenge@oncfs.gouv.fr
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 | ## Load the data:
data("dataCollision")
## Prepare the data
X <- dataCollision$RedDeer$X
## Sets the elevation <600 to NULL (redundant with other elevations:
## if the elevation is neither >1500, nor between 600 and 1500,
## it is necessarily < 600
X$elev600 <- NULL
## Set the variable roads to NULL (not useful here, as noted in the
## paper)
X$roads <- NULL
## prepares the alpha:
## vector alphas: the name of alphas is the same as the columns of X...
alphas <- names(X)
## ... except for elevation and habitat,
## which correspond to several variables in X
alphas[7:8] <- "elev"
alphas[11:14] <- "hab"
## Note that the variables Agriculture, Open, Urban, and Forest
## are strongly correlated together. To reduce the correlation and
## improve mixing, we transform these variables with the help of a PCA
## (package ade4)
hab <- X[,11:14]
X <- X[,-c(11:14)]
pc <- ade4::dudi.pca(hab,scannf = FALSE, nf=4)
## We scale the variables to improve mixing
X <- as.data.frame(scale(X))
## and we add the transformed habitat variables from the PCA:
li1 <- pc$l1
X$AX1 <- li1[,1]
X$AX2 <- li1[,2]
X$AX3 <- li1[,3]
X$AX4 <- li1[,4]
## Use of the function preparefit
pf <- prepareFit(X, alphas, dataCollision$RedDeer$coll,
dataCollision$RedDeer$Y, dataCollision$RedDeer$Area,
dataCollision$RedDeer$departement)
## Not run:
## WARNING: very long execution (about 1 hour)!!
## the results are stored in the dataset "modelRedDeer"
## But if you want to try it, the data are now ready for the fit:
mo <- jags.model(textConnection(pf$modelstring), ini=pf$ini, data=pf$data4jags)
update(mo, n.iter=1000)
modelRedDeer <- coda.samples(mo, variable.names = pf$coefnames,
n.iter = 500000, thin=100)
## End(Not run)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.