siland (version 1.4.6)

Fsiland.lik: Compute -(Log-likelihood) for models with different parameters for spatial influence functions.

Description

This function allows to investigate some problem optimization for function Fsiland by plotting the -(log-likelihhod) for models with different parameters for spatial influence functions for landscape variables.

Usage

Fsiland.lik(res, land,data, varnames = NULL, seqd = seq(2, 2000, length = 10))

Arguments

res

an object from Fsiland estimation

land

The landscape (a sf object) used to obtain the result res

data

the dataframe used to obtain the result res

varnames

landscape variables for which -(log-likelihhod) is studied. If varnames is null (by default), -(log-likelihhod) is investigated for all landscape variables in estimted model for object res.

seqd

a vector of positives values that represent different buffer sizes for the landscape variables.

Value

A matrix with -(log-likelihood) values for the different landscape variables (argument varnames) and for the different buffer sizes (argument seqd).

Details

Estimation based on log-likelihood aims to find parameters that maximises log-likelihood, that is to say that minimses -(log-likelihood). From graphics obtained with this fucntion, orange curve gives the -(log-likelihood) obtained for the estimatd model. Others curves gives -(log-likelihood) by varying buffer sizes for landscape variables. If minimisation for -(log-likelihood) for estimation given in object res, no curve should go below the ornage one.

Examples

# NOT RUN {
data(dataSiland)
data(landSiland)
resF=Fsiland(obs~x1+L1+L2,data=dataSiland,land=landSiland,init = c(50),wd=20)
Fsiland.lik(resF,dataSiland,land=landSiland,varnames=c("L1","L2"),seqd=seq(5,500,length=20))
# }