R/GWLelast.inner.R

In GWLelast: Geographically Weighted Logistic Elastic Net Regression

#' GWLelast.inner
#'
#' Inner part of fitting GWLelast without parallel cores
#'
#' @param x Covariates.
#' @param y Outcome binary variable.
#' @param coords 2 columns matrix including "longitude" and "latitude".
#' @param W Weight matrix.
#' @param alpha The elasticnet mixing parameter [0,1] in glmnet package.
#' @param lambda Optional user-supplied lambda sequence in glmnet package.
#' @param nlambda The number of lambda values in glmnet package.
#'
#' @importFrom  stats predict
#' @return model Fitted model at location i.
#' @return error Cross validation error.
#' @export


GWLelast.inner = function(x=x, y = y, coords = coords, W = W, lambda = lambda, alpha = 1, nlambda = nlambda) {

  model = list()
  cv.error = list()
  GWLelast.inner = list()

  for(i in 1:dim(x)[1]){
    w = W[i, -i]

    xi = as.matrix(x[-i,])
    yi = as.matrix(y[-i])

    model[[i]] = glmnet(x = xi, y = yi, weights = w, family = "binomial", alpha = alpha, nlambda = nlambda, lambda = lambda)
    predictions = predict(model, newx = t(x[i,]), type = "class")
    error[[i]] = abs(as.numeric(predictions) - y[i])

  }

  GWLelast.inner[["model"]] = model
  GWLelast.inner[["error"]] = error

  return(GWLelast.inner)
}