uhcdenscalc: uhcdenscalc

In aaarchmiller/uhcplots: Used-Habitat Calibration Plots (UHC Plots)

Description

uhcdenscalc calculates kernal density estimates to create UHC plots.

Usage

uhcdenscalc(rand_sims, dat, avail, gridsize = 500)

Arguments

rand_sims	The characteristics associated with the simulated locations in the test data set
dat	The characteristics associated with observed locations in the test data set
avail	The characteristics associated with available locations in the test data set
gridsize	The size of grid for estimating the kernel density estimate (kde). Default value is 500.

Details

uhcdenscalc calculates density estimates for the environmental characteristics associated with the observed and available locations in the test data and also those associated with the randomly chosen locations generated by the uhcsim or uhcsimstrat functions.

Value

A list of density estimates associated with the simulated locations, (densrand), the observed locations (densdat), and the available locations (densavail)

See Also

Full archive of the data and code necessary to replicate the manuscript at http://doi.org/10.13020/D6T590.

Examples

# Simulate training data for the non-linear example
nonlinear.train <- uhcdatasimulator(nused = 100,
    navail = 10000,
    betas = c(2,-1),
    ntemp = 1000000,
    example = "non-linear")

# Simulate test data for the non-linear example
nonlinear.test <- uhcdatasimulator(nused = 100,
    navail = 10000,
    betas = c(2,-1),
    ntemp = 1000000,
    example = "non-linear")

# Fit GLM with quadratic relationship
train.correct <- glm(y~temp + I(temp^2),
   family = binomial,
   data = nonlinear.train)

# Fit GLM with linear (misspecified) relationship
train.misspec <- glm(y~temp,
   family = binomial,
   data = nonlinear.train)

# Simulate data for quadratic model
xhat.correct <- uhcsim(nsims = 1000,
   nused_test = 100,
   xmat = model.matrix(y~temp + I(temp^2), data = nonlinear.test)[,-1],
   fit_rsf = train.correct,
   z = as.matrix(nonlinear.test[,"temp"]))

# Simulate data for linear (misspecified) model
xhat.misspec <- uhcsim(nsims = 1000,
   nused_test = 100,
   xmat = as.matrix(model.matrix(y~temp, data = nonlinear.test)[,2]),
   fit_rsf = train.misspec,
   z = as.matrix(nonlinear.test[,"temp"]))

# Get density estimates for quadratic model
denshats.correct <- uhcdenscalc(rand_sims = xhat.correct[,,1],
   dat = subset(nonlinear.test, y==1, select="temp"),
   avail = subset(nonlinear.test, y==0, select="temp"))

# Get density estimates for linear (misspecified) model
denshats.misspec <- uhcdenscalc(rand_sims = xhat.misspec[,,1],
   dat = subset(nonlinear.test, y==1, select="temp"),
   avail = subset(nonlinear.test, y==0, select="temp"))

aaarchmiller/uhcplots documentation built on May 10, 2019, 2:05 a.m.