ppmnet: Fit a spatial point process model with lasso or elastic net...
In jeffdaniel/ppmnet: Regularized Spatial Point Process Models

Description Usage Arguments Value Examples

Fits spatial point process models via penalized composite likelihood. The regularization path is computed for the lasso or elastic net penalty along a sequence of tuning parameter values. Support for Poisson point process models and for Gibbs point process models.

1 2	ppmnet(Q, data, interaction = NULL, correction = "border", rbord = NULL, method = "mpl", ...)

`Q`	A quadrature scheme (of class `quad`) containing a point pattern.
`data`	A list of pixel images (of class `imlist`) representing spatial covariates.
`interaction`	An object (of class `interact`) describing the point process interaction structure, or `NULL` indicating that a Poisson point process model is to be fit.
`correction`	Edge correction to be used. Either "`border`" for border correction, or "`none`" for no correction.
`rbord`	If `correction = "border"`, the distance by which the observation window is eroded.
`method`	Method to be used to fit the model. Either "`mpl`" for penalized maximum pseudolikelihood, or "`logi`" for penalized maximum logistic composite likelihood.
`...`	Additional arguments passed to `glmnet` to control the fitting procedure.

An object of class "ppmnet".

# Poisson model fit via penalized maximum likelihood
Qp <- quadscheme(Xp)
fitp <- ppmnet(Qp, exdata)

# Strauss model fit via penalized maximum pseudolikelihood
Qs <- quadscheme(Xs)
fits <- ppmnet(Qs, exdata, Strauss(5), nlambda = 20)

# Geyer saturation model fit via penalized logistic composite likelihood
Qg <- quadscheme.logi(Xg)
fitg <- ppmnet(Qg, exdata, Geyer(5, 1), method = "logi", nlambda = 20)