lmeHyperframe: Fits a set of mixed effects models to a hyperframe object

In robertbagchi/ReplicatedPointPatterns: Analysis of replicated point patterns with covariates

Description

Fits a mixed effects model to a given hyperframe object with K function (K) and weights (wts) elements.

kfunclme and refit.lmek are internally called by lmeHyperframe.

Usage

lmeHyperframe(hyper, r, fixed, random, correlation = NULL, correction =
"border", weights.type, computeK=TRUE, minsamp = NA, printwarnings =
TRUE)

kfuncLme(k, dat, weights, fixed, random, correlation, na.action = na.omit)
refit.lmek(mod, res.r)

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

Arguments

hyper	Hyperframe object created with spatstat(as.hyperframe). Must K and wts elements.
r	Distances at which to model K. Should include 0
fixed	RHS of the fixed effects formula in quotes, without a tilda (e.g. "x1 + x2")
random	RHS of the random effects formula in quotes (e.g. "1|group")
correlation	The correlation argument for models - sent to nlme::lme and must be accepted by nlme::lme.
correction	The edge correction which must correspond to an option in spatstat::Kest
weights.type	The regression weights. Must be one of "nx", "nx_A", "nx2", "nx2_A", "sqrtnxny", "nxny", "nxny_A", "sqrtnxny_A"
computeK	Should K be computed internally. Defaults to TRUE
minsamp	The minimum number of points, after edge corrections, for a case to be included. Cases with fewer points will be excluded and row names returned as an attribute.
printwarnings	Should warnings about case or distance exclusion be reported. Defaults to TRUE
k	A vector corresponding the K(r) for all cases at a given distance r
dat	Data including the covariates used in the fixed and random parts of the model.
weights	The model weights passed to lme.
na.action	How to deal with NAs in the data set. Defaults to na.omit.
mod	An lme model at each distance. Used internally by refit.lmek
res.r	Vector of residuals. Used internally by refit.lmek
x	Object to be printed.

Author(s)

Robert Bagchi Maintainer: Robert Bagchi <[e-mail]>

Examples

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (hyper, r, fixed, random, correlation = NULL, correction = "border", 
    weights.type, minsamp = 10, printwarnings = TRUE) 
{
    if (min(r) > 0) 
        r <- c(0, r)
    if (!all(c("pppx") %in% names(hyper))) {
        stop("hyperframe object must include 'pppx' element")
    }
    hyper$K <- with.hyperframe(hyper, Kest(pppx, r = r, correction = correction, 
        ratio = TRUE))
    hyper$wts <- with.hyperframe(hyper, list(kfunc.weights.calc(pppx, 
        r = K$r, correction = correction, type = weights.type)))
    sp.keep <- sapply(with.hyperframe(hyper, list(kfunc.weights.calc(pppx, 
        r = K$r, correction = correction, type = "nx"))), function(x) all(unlist(x[r]) >= 
        minsamp))
    removed.species <- row.names(hyper)[!sp.keep]
    if (printwarnings) 
        warning(paste("Removed", length(removed.species), "species with insufficient numbers"))
    hyper <- hyper[sp.keep, ]
    dist.keep <- (apply(sapply(hyper$K, function(K) K[[correction]]), 
        1, function(x) var(x)) > 0)
    if (printwarnings) 
        warning(paste("Not modelling K at distances ", paste(r[!dist.keep], 
            collapse = ", "), "due to zero variance"))
    modr <- match(r[dist.keep], r)
    kmods <- sapply(modr, function(i) {
        kfuncLme(k = sapply(hyper$K, function(k) k[[correction]][i]), 
            dat = as.data.frame(hyper, warn = FALSE), fixed = fixed, 
            random = random, correlation = correlation, weights = sapply(hyper$wts, 
                function(w) unlist(w)[i]))
    }, simplify = FALSE)
    names(kmods) <- r[dist.keep]
    return(kmods)
  }

robertbagchi/ReplicatedPointPatterns documentation built on May 27, 2019, 10:32 a.m.