ppmlasso-package: PPM-LASSO: Point process models with LASSO-type penalties

ppmlasso-packageR Documentation

PPM-LASSO: Point process models with LASSO-type penalties

Description

This package contains tools to fit point process models with sequences of LASSO penalties ("regularisation paths"). Regularisation paths of Poisson point process models or area-interaction models can be fitted with LASSO, adaptive LASSO or elastic net penalties. A number of criteria are available to judge the bias-variance tradeoff.

Details

The key functions in ppmlasso are as follows:

Useful pre-analysis functions:

findRes

Determine the optimal spatial resolution at which to perform analysis

getEnvVar

Interpolate environmental data to species presence locations

griddify

Ensure a matrix of environmental data is on a rectangular grid

ppmdat

Calculate observation weights and set up design matrix for fitting

pointInteractions

Calculate interpoint interactions for fitting area-interaction models

sampleQuad

Set up a regular grid of quadrature points

Creating regularisation paths of point process models:

ppmlasso

Fit a regularisation path of point process models

plotFit

Plot the fitted intensity of a ppmlasso object

plotPath

Plot the regularisation path of a ppmlasso object

print.ppmlasso

Print output from a ppmlasso object

predict.ppmlasso

Make predictions from a fitted point process model to new data

Checking assumptions:

diagnose.ppmlasso

Create diagnostic residual plots of ppmlasso object

envelope.ppmlasso

Create simulation envelope for goodness-of-fit checks on a ppmlasso object

Author(s)

Ian W. Renner

Maintainer: Ian W. Renner <Ian.Renner@newcastle.edu.au>

References

Renner, I.W. & Warton, D.I. (2013). Equivalence of MAXENT and Poisson point process models for species distribution modeling in ecology Biometrics 69, 274-281.

Renner, I.W. et al (2015). Point process models for presence-only analysis. Methods in Ecology and Evolution 6, 366-379.

Renner, I.W., Warton, D.I., & Hui, F.K.C. (2021). What is the effective sample size of a spatial point process? Australian & New Zealand Journal of Statistics 63, 144-158.

Examples

# Fit a regularisation path of Poisson point process models
data(BlueMountains)
sub.env = BlueMountains$env[BlueMountains$env$Y > 6270 & BlueMountains$env$X > 300,]
sub.euc = BlueMountains$eucalypt[BlueMountains$eucalypt$Y > 6270 & BlueMountains$eucalypt$X > 300,]
ppm.form = ~ poly(FC, TMP_MIN, TMP_MAX, RAIN_ANN, degree = 2, raw = TRUE)
ppm.fit  = ppmlasso(ppm.form, sp.xy = sub.euc, env.grid = sub.env, sp.scale = 1, n.fits = 20,
writefile = FALSE)

# Fit a regularisation path of area-interaction models
data(BlueMountains)
ai.form  = ~ poly(FC, TMP_MIN, TMP_MAX, RAIN_ANN, degree = 2, raw = TRUE)
ai.fit   = ppmlasso(ai.form, sp.xy = sub.euc, env.grid = sub.env, sp.scale = 1, 
family = "area.inter", r = 2, availability = BlueMountains$availability, n.fits = 20,
writefile = FALSE)

# Print a ppmlasso object
print(ppm.fit, out = "model")

# Residual plot of a ppmlasso object
diagnose(ppm.fit, which = "smooth", type = "Pearson")

# Make predictions
pred.mu = predict(ppm.fit, newdata = sub.env)

# Plot the intensity from a fitted ppmlasso object
plotFit(ppm.fit)

# Plot the regularisation path from a fitted ppmlasso object
plotPath(ppm.fit)


ppmlasso documentation built on May 29, 2024, 7:12 a.m.