R/pnp_lobagoc.R

In S4DM: Small Sample Size Species Distribution Modeling

#' @name pnp_lobagoc
#' @title Internal function for fitting lobagoc distributions in plug-and-play SDMs.
#' @description This function both fits lobagoc distributions \insertCite{Drake2014-qz}{S4DM} and projects those distributions to new covariates.
#' @param data dataframe of covariates
#' @param method one of either "fit" or "predict"
#' @param object fitted object returned by a pnp_... function. Only needed when method = "predict"
#' @param v Positive integer. The Number of votes to use (default is 100)
#' @param nu Numeric. Tuning parameter for nu-svm
#' @param sigma NULL or Numeric.  Tuning parameter of rbf kernel, will estimate if left NULL (default).
#' @details For fitting, an object is not required (and will be ignored). For prediction, parameters v,nu,and sigma are not needed and will be ignored.
#' @importFrom  kernlab ksvm predict
#' @keywords internal
#' @references
#' \insertAllCited{}
pnp_lobagoc <- function(data, method, object = NULL, v = 100, nu = 0.01, sigma = NULL){

  #Code to check inputs

  #Code for fitting
  if(method == "fit"){

    labels <- c(rep(1, dim(data)[1]))
    models <- list()
    for(i in 1:v){
      data_i <- rbind(data[sample.int(n = dim(data)[1],
                                 replace = TRUE),])
      if(!is.null(sigma)){

        models[[i]] <- ksvm(labels~.,
                            data = cbind(data, labels),
                            type='one-svc',
                            nu=nu,
                            kpar=list(sigma = sigma))
      }
      else{

        models[[i]] <- ksvm(labels~.,
                            data = cbind(data, labels),
                            type='one-svc',
                            nu = nu)
      }
    }


    model <- list(lobagoc_models = models,
                  method = "lobagoc")

    class(model) <- "pnp_estimate"
    return(model)

  }

  #Code for predicting

  if(method == "predict"){

    v <- length(object$lobagoc_models)
    prediction <- data[,1]*0

    for(i in 1:v){
      prediction <- prediction + predict(object$lobagoc_models[[i]], data)
    }

    prediction <- (prediction/v)

    return(log(prediction))
  }

}