R/vif.ev.R

In spfilteR: Semiparametric Spatial Filtering with Eigenvectors in (Generalized) Linear Models

#' @name vif.ev
#'
#' @title Variance Inflation Factor of Eigenvectors
#'
#' @description Calculate the variance inflation factor (VIF) of
#' the eigenvectors in the spatial filter.
#'
#' @param x vector/ matrix of regressors (default = NULL)
#' @param evecs (selected) eigenvectors
#' @param na.rm remove missing values in covariates (TRUE/ FALSE)
#'
#' @return Returns a vector containing the VIF for each eigenvector.
#'
#' @note This function assumes a linear model which ensures the
#' uncorrelatedness of the eigenvectors. Note that regression weights
#' or the link function used in generalized linear models can corrupt
#' this property.
#'
#' @examples
#' data(fakedata)
#' E <- getEVs(W = W, covars = NULL)$vectors
#' (VIF <- vif.ev(x = fakedataset$x1, evecs = E[, 1:10]))
#'
#' @importFrom stats complete.cases
#'
#' @author Sebastian Juhl
#'
#' @seealso \code{\link{lmFilter}}, \code{\link{getEVs}}
#'
#' @export

vif.ev <- function(x = NULL, evecs, na.rm = TRUE) {
  evecs <- as.matrix(evecs)
  if (is.null(x)) {
    x <- rep(1, nrow(evecs))
  }
  x <- as.matrix(x)
  if (!isTRUE(all.equal(x[, 1], rep(1, nrow(evecs))))) {
    x <- cbind(1, x)
  }
  if (na.rm) {
    evecs <- as.matrix(evecs[complete.cases(x),])
    x <- as.matrix(x[complete.cases(x),])
  }
  if (anyNA(x)) {
    stop("Missing values detected")
  }
  inflate <- NULL
  for (i in seq_len(ncol(evecs))) {
    xi <- evecs[, i]
    tots <- sum((xi - mean(xi))^2)
    ests <- solve(crossprod(x)) %*% crossprod(x, xi)
    re <- xi - x %*% ests
    Rsq <- 1 - (sum(crossprod(re)) / tots)
    inflate[i] <- 1 / (1 - Rsq)
  }
  return(inflate)
}