AIC

Description

Akaike information criterion of a moveHMM model.

Usage

## S3 method for class 'moveHMM'
AIC(object, ..., k = 2)

Arguments

Value

The AIC of the model(s) provided. If several models are provided, the AICs are output in ascending order.

Examples

# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m
AIC(m)

Confidence intervals

Description

Computes the confidence intervals of the step length and turning angle parameters, as well as for the transition probabilities regression parameters.

Usage

CI(m, alpha = 0.95, nbSims = 10^6)

Arguments

Value

A list of the following objects:

Examples

# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m

CI(m)

Matrix of all probabilities

Description

Used in functions viterbi, logAlpha, logBeta.

Usage

allProbs(
  data,
  nbStates,
  stepDist,
  angleDist,
  stepPar,
  anglePar = NULL,
  zeroInflation = FALSE,
  knownStates = NULL
)

Arguments

Value

Matrix of all probabilities.

Examples

## Not run: 
stepPar <- c(1,10,1,5,0.2,0.3)
anglePar <- c(0,pi,0.5,2)
stepDist <- "gamma"
angleDist <- "vm"
data <- simData(nbAnimals=5,nbStates=2,stepDist=stepDist,angleDist=angleDist,stepPar=stepPar,
                 anglePar=anglePar,nbCovs=2,zeroInflation=TRUE)
P <- allProbs(data=data,nbStates=2,stepDist=stepDist,angleDist=angleDist,
               stepPar=matrix(stepPar,ncol=2,byrow=TRUE),anglePar=matrix(anglePar,ncol=2,
               byrow=TRUE),zeroInflation=TRUE)

## End(Not run)

Confidence intervals for angle parameters

Description

Simulation-based computation of confidence intervals for the parameters of the angle distribution. Used in CI.

Usage

angleCI(m, alpha, nbSims = 10^6)

Arguments

Value

A list of the following objects:

Exponential density function

Description

Probability density function of the exponential distribution (written in C++)

Usage

dexp_rcpp(x, rate, foo = 0)

Arguments

Value

Vector of densities

Gamma density function

Description

Probability density function of the gamma distribution (written in C++)

Usage

dgamma_rcpp(x, mu, sigma)

Arguments

Value

Vector of densities

Log-normal density function

Description

Probability density function of the log-normal distribution (written in C++)

Usage

dlnorm_rcpp(x, meanlog, sdlog)

Arguments

Value

Vector of densities

Density function of von Mises distribution

Description

Density function of von Mises distribution

Usage

dvm(x, mu, kappa, log = FALSE)

Arguments

Value

Von Mises density

Von Mises density function

Description

Probability density function of the Von Mises distribution, defined as a function of the modified Bessel function of order 0 (written in C++)

Usage

dvm_rcpp(x, mu, kappa)

Arguments

Value

Vector of densities

Weibull density function

Description

Probability density function of the Weibull distribution (written in C++)

Usage

dweibull_rcpp(x, shape, scale)

Arguments

Value

Vector of densities

Density function of wrapped Cauchy distribution

Description

Density function of wrapped Cauchy distribution

Usage

dwrpcauchy(x, mu, rho, log = FALSE)

Arguments

Value

Wrapped Cauchy density

Wrapped Cauchy density function

Description

Probability density function of the wrapped Cauchy distribution (written in C++)

Usage

dwrpcauchy_rcpp(x, mu, rho)

Arguments

Value

Vector of densities

Elk data set from Morales et al. (2004, Ecology)

Description

It is a data frame with the following columns:

• ID Track identifier

• Easting Easting coordinate of locations

• Northing Northing coordinate of locations

• dist_water Distance of elk to water (in metres)

Usage

elk_data

Example data simulation

Description

Generate the file data/example.RData, used in other functions' examples and unit tests.

Usage

exGen()

Example dataset

Description

This data is generated by the function exGen, and used in the examples and tests of other functions to keep them as short as possible.

Usage

example

Details

It is a list of the following objects:

• data A moveData object

• m A moveHMM object

• simPar The parameters used to simulate data

• par0 The initial parameters in the optimization to fit m

Fit an HMM to the data

Description

Fit an hidden Markov model to the data provided, using numerical optimization of the log-likelihood function.

Usage

fitHMM(
  data,
  nbStates,
  stepPar0,
  anglePar0 = NULL,
  beta0 = NULL,
  delta0 = NULL,
  formula = ~1,
  stepDist = c("gamma", "weibull", "lnorm", "exp"),
  angleDist = c("vm", "wrpcauchy", "none"),
  angleMean = NULL,
  stationary = FALSE,
  knownStates = NULL,
  verbose = 0,
  nlmPar = NULL,
  fit = TRUE
)

Arguments

Details

• The matrix beta of regression coefficients for the transition probabilities has one row for the intercept, plus one row for each covariate, and one column for each non-diagonal element of the transition probability matrix. For example, in a 3-state HMM with 2 covariates, the matrix beta has three rows (intercept + two covariates) and six columns (six non-diagonal elements in the 3x3 transition probability matrix - filled in row-wise). In a covariate-free model (default), beta has one row, for the intercept.

• The choice of initial parameters is crucial to fit a model. The algorithm might not find the global optimum of the likelihood function if the initial parameters are poorly chosen.

Value

A moveHMM object, i.e. a list of:

References

Patterson T.A., Basson M., Bravington M.V., Gunn J.S. 2009. Classifying movement behaviour in relation to environmental conditions using hidden Markov models. Journal of Animal Ecology, 78 (6), 1113-1123.

Langrock R., King R., Matthiopoulos J., Thomas L., Fortin D., Morales J.M. 2012. Flexible and practical modeling of animal telemetry data: hidden Markov models and extensions. Ecology, 93 (11), 2336-2342.

Examples

set.seed(872)
### 1. simulate data
# define all the arguments of simData
nbAnimals <- 2
nbStates <- 2
nbCovs <- 2
mu<-c(15,50)
sigma<-c(10,20)
angleMean <- c(pi,0)
kappa <- c(0.7,1.5)
stepPar <- c(mu,sigma)
anglePar <- c(angleMean,kappa)
stepDist <- "gamma"
angleDist <- "vm"
zeroInflation <- FALSE
obsPerAnimal <- c(50,100)

data <- simData(nbAnimals=nbAnimals,nbStates=nbStates,stepDist=stepDist,angleDist=angleDist,
                 stepPar=stepPar,anglePar=anglePar,nbCovs=nbCovs,zeroInflation=zeroInflation,
                 obsPerAnimal=obsPerAnimal)

### 2. fit the model to the simulated data
# define initial values for the parameters
mu0 <- c(20,70)
sigma0 <- c(10,30)
kappa0 <- c(1,1)
stepPar0 <- c(mu0,sigma0) # no zero-inflation, so no zero-mass included
anglePar0 <- kappa0 # the angle mean is not estimated, so only the concentration parameter is needed
formula <- ~cov1+cos(cov2)

m <- fitHMM(data=data,nbStates=nbStates,stepPar0=stepPar0,anglePar0=anglePar0,formula=formula,
              stepDist=stepDist,angleDist=angleDist,angleMean=angleMean)

print(m)

Discrete colour palette for states

Description

Discrete colour palette for states

Usage

getPalette(nbStates)

Arguments

Value

Vector of colours, of length nbStates.

Data to produce plots of fitted model

Description

Data to produce plots of fitted model

Usage

getPlotData(m, type, format = "wide", alpha = 0.95)

Arguments

Details

• If type = "dist", the function evaluates each state-dependent distribution over the range of observed variable (step length or turning angle), and weighs them by the proportion of time spent in each state (obtained from Viterbi state sequence).

• If type = "tpm", the function returns transition probabilities estimated over a range of covariate values. Other covariates are fixed to their mean values.

Value

Data frame (or list of data frames) containing data in a format that can easily be plotted. If type = "dist", the output is a list with two elements, "step" and "angle". If type = "tpm" or "stat", the output is a list with one element for each covariate. See details for more extensive description of output.

Wild haggis data set from Michelot et al. (2016, Methods Eco Evol)

Description

Data frame of the first three tracks from Michelot et al. (2016), with columns:

• ID Track identifier

• x Easting coordinate of locations

• y Northing coordinate of locations

• slope Terrain slope (in degrees)

• temp Air temperature (in degrees Celsius)

Usage

haggis_data

Is moveData

Description

Check that an object is of class moveData. Used in fitHMM.

Usage

is.moveData(x)

Arguments

Value

TRUE if x is of class moveData, FALSE otherwise.

Is moveHMM

Description

Check that an object is of class moveHMM. Used in CI, plotPR, plotStates, pseudoRes, stateProbs, and viterbi.

Usage

is.moveHMM(x)

Arguments

Value

TRUE if x is of class moveHMM, FALSE otherwise.

Forward log-probabilities

Description

Used in stateProbs and pseudoRes.

Usage

logAlpha(m)

Arguments

Value

The matrix of forward log-probabilities.

Examples

## Not run: 
# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m

la <- logAlpha(m)

## End(Not run)

Backward log-probabilities

Description

Used in stateProbs.

Usage

logBeta(m)

Arguments

Value

The matrix of backward log-probabilities.

Examples

## Not run: 
# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m

lb <- logBeta(m)

## End(Not run)

Constructor of moveData objects

Description

This constructor is not intended to be used, except inside the function prepData. Refer to the documentation for that function.

Usage

moveData(data)

Arguments

Value

An object of class moveData.

Constructor of moveHMM objects

Description

This constructor is not intended to be used, except inside the function fitHMM. Refer to the documentation for that function.

Usage

moveHMM(m)

Arguments

Value

An object moveHMM.

Scaling function: natural to working parameters.

Description

Scales each parameter from its natural interval to the set of real numbers, to allow for unconstrained optimization. Used during the optimization of the log-likelihood.

Usage

n2w(par, bounds, beta, delta = NULL, nbStates, estAngleMean)

Arguments

Value

A vector of unconstrained parameters.

Examples

## Not run: 
nbStates <- 3
par <- c(0.001,0.999,0.5,0.001,1500.3,7.1)
bounds <- matrix(c(0,1, # bounds for first parameter
                   0,1, # bounds for second parameter
                   0,1, # ...
                   0,Inf,
                   0,Inf,
                   0,Inf),
                 byrow=TRUE,ncol=2)
beta <- matrix(rnorm(18),ncol=6,nrow=3)
delta <- c(0.6,0.3,0.1)

# vector of working parameters
wpar <- n2w(par=par,bounds=bounds,beta=beta,delta=delta,nbStates=nbStates,
           estAngleMean=FALSE)

## End(Not run)

Negative log-likelihood function

Description

Negative log-likelihood function

Usage

nLogLike(
  wpar,
  nbStates,
  bounds,
  parSize,
  data,
  covs,
  stepDist = c("gamma", "weibull", "lnorm", "exp"),
  angleDist = c("vm", "wrpcauchy", "none"),
  angleMean = NULL,
  zeroInflation = FALSE,
  stationary = FALSE,
  knownStates = NULL
)

Arguments

Value

The negative log-likelihood of the parameters given the data.

Examples

## Not run: 
# data is a moveData object (as returned by prepData), automatically loaded with the package
simPar <- example$simPar
par0 <- example$par0

estAngleMean <- is.null(simPar$angleMean)
bounds <- moveHMM:::parDef(simPar$stepDist,simPar$angleDist,simPar$nbStates,
                           estAngleMean,simPar$zeroInflation)$bounds
parSize <- moveHMM:::parDef(simPar$stepDist,simPar$angleDist,simPar$nbStates,
                            estAngleMean,simPar$zeroInflation)$parSize

par <- c(par0$stepPar0,par0$anglePar0)
wpar <- moveHMM:::n2w(par,bounds,par0$beta0,par0$delta0,simPar$nbStates,FALSE)

covs <- model.matrix(example$m$conditions$formula, example$data)

l <- nLogLike(wpar=wpar,nbStates=simPar$nbStates,bounds=bounds,parSize=parSize,
              data=data,covs=covs, stepDist=simPar$stepDist,
              angleDist=simPar$angleDist,
              zeroInflation=simPar$zeroInflation)

## End(Not run)

Negative log-likelihood

Description

Computation of the negative log-likelihood (forward algorithm - written in C++)

Usage

nLogLike_rcpp(
  nbStates,
  beta,
  covs,
  data,
  stepDist,
  angleDist,
  stepPar,
  anglePar,
  delta,
  aInd,
  zeroInflation,
  stationary,
  knownStates
)

Arguments

Value

Negative log-likelihood

Parameters definition

Description

Parameters definition

Usage

parDef(stepDist, angleDist, nbStates, estAngleMean, zeroInflation)

Arguments

Value

A list of:

Plot moveData

Description

Plot moveData

Usage

## S3 method for class 'moveData'
plot(x, animals = NULL, compact = FALSE, ask = TRUE, breaks = "Sturges", ...)

Arguments

Examples

# data is a moveData object (as returned by prepData), automatically loaded with the package
data <- example$data

plot(data,compact=TRUE,breaks=20,ask=FALSE)

Plot moveHMM

Description

Plot the fitted step and angle densities over histograms of the data, transition probabilities as functions of the covariates, and maps of the animals' tracks colored by the decoded states.

Usage

## S3 method for class 'moveHMM'
plot(
  x,
  animals = NULL,
  ask = TRUE,
  breaks = "Sturges",
  col = NULL,
  plotTracks = TRUE,
  plotCI = FALSE,
  alpha = 0.95,
  ...
)

Arguments

Details

The state-dependent densities are weighted by the frequency of each state in the most probable state sequence (decoded with the function viterbi). For example, if the most probable state sequence indicates that one third of observations correspond to the first state, and two thirds to the second state, the plots of the densities in the first state are weighted by a factor 1/3, and in the second state by a factor 2/3.

Examples

# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m

plot(m,ask=TRUE,animals=1,breaks=20)

Plot pseudo-residuals

Description

Plots time series, qq-plots (against the standard normal distribution), and sample ACF functions of the pseudo-residuals

Usage

plotPR(m)

Arguments

Details

• If some turning angles in the data are equal to pi, the corresponding pseudo-residuals will not be included. Indeed, given that the turning angles are defined on (-pi,pi], an angle of pi results in a pseudo-residual of +Inf (check Section 6.2 of reference for more information on the computation of pseudo-residuals).

• If some steps are of length zero (i.e. if there is zero-inflation), the corresponding pseudo- residuals are shown as segments, because pseudo-residuals for discrete data are defined as segments (see Zucchini and MacDonald, 2009, Section 6.2).

References

Zucchini, W. and MacDonald, I.L. 2009. Hidden Markov Models for Time Series: An Introduction Using R. Chapman & Hall (London).

Examples

# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m

plotPR(m)

Plot observations on satellite image

Description

Plot tracking data on a satellite map. This function only works with longitude and latitude values (not with UTM coordinates), and uses the package ggmap to fetch a satellite image from Google. An Internet connection is required to use this function.

Usage

plotSat(
  data,
  zoom = NULL,
  location = NULL,
  segments = TRUE,
  compact = TRUE,
  col = NULL,
  alpha = 1,
  size = 1,
  states = NULL,
  animals = NULL,
  ask = TRUE,
  return = FALSE
)

Arguments

Details

If the plot displays the message "Sorry, we have no imagery here", try a lower level of zoom.

References

D. Kahle and H. Wickham. ggmap: Spatial Visualization with ggplot2. The R Journal, 5(1), 144-161. URL: http://journal.r-project.org/archive/2013-1/kahle-wickham.pdf

Plot states

Description

Plot the states and states probabilities.

Usage

plotStates(m, animals = NULL, ask = TRUE)

Arguments

Examples

# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m

# plot states for first and second animals
plotStates(m,animals=c(1,2))

Plot stationary state probabilities

Description

Plot stationary state probabilities

Usage

plotStationary(m, col = NULL, plotCI = FALSE, alpha = 0.95)

Arguments

Examples

# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m

plotStationary(m)

Predict stationary state probabilities

Description

Predict stationary state probabilities

Usage

predictStationary(
  m,
  newData,
  beta = m$mle$beta,
  returnCI = FALSE,
  alpha = 0.95
)

Arguments

Value

List with elements 'mle', 'lci', and 'uci' (the last two only if returnCI = TRUE). Each element is a matrix of stationary state probabilities with one row for each row of newData and one column for each state.

Predict transition probabilities for new covariate values

Description

Predict transition probabilities for new covariate values

Usage

predictTPM(m, newData, beta = m$mle$beta, returnCI = FALSE, alpha = 0.95)

Arguments

Value

List with elements 'mle', 'lci', and 'uci' (the last two only if returnCI = TRUE). Each element is an array, where each layer is a transition probability matrix corresponding to a row of newData.

Preprocessing of the tracking data

Description

Preprocessing of the tracking data

Usage

prepData(
  trackData,
  type = c("LL", "UTM"),
  coordNames = c("x", "y"),
  LLangle = NULL
)

Arguments

Value

An object moveData, i.e. a dataframe of:

Examples

coord1 <- c(1,2,3,4,5,6,7,8,9,10)
coord2 <- c(1,1,1,2,2,2,1,1,1,2)
trackData <- data.frame(coord1=coord1,coord2=coord2)
d <- prepData(trackData,type='UTM',coordNames=c("coord1","coord2"))

Print moveHMM

Description

Print moveHMM

Usage

## S3 method for class 'moveHMM'
print(x, ...)

Arguments

Examples

# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m

print(m)

Pseudo-residuals

Description

The pseudo-residuals of a moveHMM model, as described in Zucchini and McDonad (2009).

Usage

pseudoRes(m)

Arguments

Details

If some turning angles in the data are equal to pi, the corresponding pseudo-residuals will not be included. Indeed, given that the turning angles are defined on (-pi,pi], an angle of pi results in a pseudo-residual of +Inf (check Section 6.2 of reference for more information on the computation of pseudo-residuals).

Value

A list of:

References

Zucchini, W. and MacDonald, I.L. 2009. Hidden Markov Models for Time Series: An Introduction Using R. Chapman & Hall (London).

Examples

# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m
res <- pseudoRes(m)
qqnorm(res$stepRes)
qqnorm(res$angleRes)

Sample from von Mises distribution

Description

Sample from von Mises distribution

Usage

rvm(n, mu, kappa)

Arguments

Details

Uses basic rejection sampling, based on dvm(), which might be inefficient for large kappa. Could be improved following Best & Fisher (1979), Efficient simulation of the von Mises distribution, JRSSC, 28(2), 152-157.

Value

Vector of n samples from vm(mu, kappa)

Sample from wrapped Cauchy distribution

Description

Sample from wrapped Cauchy distribution

Usage

rwrpcauchy(n, mu, rho)

Arguments

Details

Uses basic rejection sampling, based on dwrpcauchy(), which might be inefficient for large rho.

Value

Vector of n samples from wrpcauchy(mu, rho)

Simulation tool

Description

Simulates movement data from an HMM.

Usage

simData(
  nbAnimals = 1,
  nbStates = 2,
  stepDist = c("gamma", "weibull", "lnorm", "exp"),
  angleDist = c("vm", "wrpcauchy", "none"),
  stepPar = NULL,
  anglePar = NULL,
  beta = NULL,
  covs = NULL,
  nbCovs = 0,
  zeroInflation = FALSE,
  obsPerAnimal = c(500, 1500),
  model = NULL,
  states = FALSE
)

Arguments

Details

• The matrix beta of regression coefficients for the transition probabilities has one row for the intercept, plus one row for each covariate, and one column for each non-diagonal element of the transition probability matrix. For example, in a 3-state HMM with 2 covariates, the matrix beta has three rows (intercept + two covariates) and six columns (six non-diagonal elements in the 3x3 transition probability matrix - filled in row-wise). In a covariate-free model (default), beta has one row, for the intercept.

• If the length of covariate values passed (either through 'covs', or 'model') is not the same as the number of observations suggested by 'nbAnimals' and 'obsPerAnimal', then the series of covariates is either shortened (removing last values - if too long) or extended (starting over from the first values - if too short).

Value

An object moveData, i.e. a dataframe of:

Examples

# 1. Pass a fitted model to simulate from
# (m is a moveHMM object - as returned by fitHMM - automatically loaded with the package)
# We keep the default nbAnimals=1.
m <- example$m
obsPerAnimal=c(50,100)
data <- simData(model=m,obsPerAnimal=obsPerAnimal)

# 2. Pass the parameters of the model to simulate from
stepPar <- c(1,10,1,5,0.2,0.3) # mean1, mean2, sd1, sd2, z1, z2
anglePar <- c(pi,0,0.5,2) # mean1, mean2, k1, k2
stepDist <- "gamma"
angleDist <- "vm"
data <- simData(nbAnimals=5,nbStates=2,stepDist=stepDist,angleDist=angleDist,stepPar=stepPar,
               anglePar=anglePar,nbCovs=2,zeroInflation=TRUE,obsPerAnimal=obsPerAnimal)

stepPar <- c(1,10,1,5) # mean1, mean2, sd1, sd2
anglePar <- c(pi,0,0.5,0.7) # mean1, mean2, k1, k2
stepDist <- "weibull"
angleDist <- "wrpcauchy"
data <- simData(nbAnimals=5,nbStates=2,stepDist=stepDist,angleDist=angleDist,stepPar=stepPar,
               anglePar=anglePar,obsPerAnimal=obsPerAnimal)

# step length only and zero-inflation
stepPar <- c(1,10,1,5,0.2,0.3) # mean1, mean2, sd1, sd2, z1, z2
stepDist <- "gamma"
data <- simData(nbAnimals=5,nbStates=2,stepDist=stepDist,angleDist="none",stepPar=stepPar,
               nbCovs=2,zeroInflation=TRUE,obsPerAnimal=obsPerAnimal)

# include covariates
# (note that it is useless to specify "nbCovs", which is determined
# by the number of columns of "cov")
cov <- data.frame(temp=rnorm(500,20,5))
stepPar <- c(1,10,1,5) # mean1, mean2, sd1, sd2
anglePar <- c(pi,0,0.5,2) # mean1, mean2, k1, k2
stepDist <- "gamma"
angleDist <- "vm"
data <- simData(nbAnimals=5,nbStates=2,stepDist=stepDist,angleDist=angleDist,stepPar=stepPar,
                anglePar=anglePar,covs=cov)

Split track at gaps

Description

Defines a new ID variable, which changes where there is a long gap in the data. This is sometimes useful for preprocessing data prior to using prepData.

Usage

splitAtGaps(data, maxGap = 60, shortestTrack = 0, units = "mins")

Arguments

Value

Data frame with identical structure as input, where ID column has been replaced by new ID for split tracks. Old ID still accessible as ID_old column

State probabilities

Description

For a given model, computes the probability of the process being in the different states at each time point.

Usage

stateProbs(m)

Arguments

Value

The matrix of state probabilities, with element [i,j] the probability of being in state j in observation i.

References

Zucchini, W. and MacDonald, I.L. 2009. Hidden Markov Models for Time Series: An Introduction Using R. Chapman & Hall (London).

Examples

# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m

sp <- stateProbs(m)

Stationary state probabilities

Description

Calculates the stationary probabilities of each state, for given covariate values.

Usage

stationary(m, covs, beta = m$mle$beta)

Arguments

Value

Matrix of stationary state probabilities. Each row corresponds to a row of covs, and each column corresponds to a state.

Examples

# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m

# data frame of covariates
stationary(m, covs = data.frame(cov1 = 0, cov2 = 0))

# design matrix (each column corresponds to row of m$mle$beta)
stationary(m, covs = matrix(c(1,0,cos(0)),1,3))

Summary moveData

Description

Summary moveData

Usage

## S3 method for class 'moveData'
summary(object, ...)

Arguments

Examples

# Use example data set as output by prepData
summary(example$data)

Transition probability matrix

Description

Computation of the transition probability matrix, as a function of the covariates and the regression parameters. Written in C++. Used in fitHMM, logAlpha, logBeta, plot.moveHMM, pseudoRes, and viterbi.

Usage

trMatrix_rcpp(nbStates, beta, covs)

Arguments

Value

Three dimensional array trMat, such that trMat[,,t] is the transition matrix at time t.

Turning angle

Description

Used in prepData.

Usage

turnAngle(x, y, z, LLangle)

Arguments

Value

The angle between vectors (x,y) and (y,z)

Examples

## Not run: 
x <- c(0,0)
y <- c(4,6)
z <- c(10,7)
turnAngle(x,y,z,LLangle=FALSE)

## End(Not run)

Viterbi algorithm

Description

For a given model, reconstructs the most probable states sequence, using the Viterbi algorithm.

Usage

viterbi(m, newdata = NULL)

Arguments

Value

The sequence of most probable states.

References

Zucchini, W. and MacDonald, I.L. 2009. Hidden Markov Models for Time Series: An Introduction Using R. Chapman & Hall (London).

Examples

# m is a moveHMM object (as returned by fitHMM), automatically loaded with the package
m <- example$m

# reconstruction of states sequence
states <- viterbi(m)

Scaling function: working to natural parameters

Description

Scales each parameter from the set of real numbers, back to its natural interval. Used during the optimization of the log-likelihood.

Usage

w2n(wpar, bounds, parSize, nbStates, nbCovs, estAngleMean, stationary)

Arguments

Value

A list of:

Examples

## Not run: 
nbStates <- 3
nbCovs <- 2
par <- c(0.001,0.999,0.5,0.001,1500.3,7.1)
parSize <- c(1,1)
bounds <- matrix(c(0,1,0,1,0,1,
                   0,Inf,0,Inf,0,Inf),
                 byrow=TRUE,ncol=2)
beta <- matrix(rnorm(18),ncol=6,nrow=3)
delta <- c(0.6,0.3,0.1)
wpar <- n2w(par,bounds,beta,delta,nbStates,FALSE)
print(w2n(wpar,bounds,parSize,nbStates,nbCovs,estAngleMean=FALSE,stationary=FALSE))

## End(Not run)