R/s_QRNN.R

In egenn/rtemis: Machine Learning and Visualization

# s_QRNN.R
# ::rtemis::
# 2015 E.D. Gennatas www.lambdamd.org
#
# Description
#   Train a quantile regression neural network and validate it.
#
# Notes
#   Can replace qrnn.nlm to add resample support instead of regular bootstrap

#' Quantile Regression Neural Network \[R\]
#'
#' Train an ensemble of Neural Networks to perform Quantile Regression using `qrnn`
#'
#' For more details on hyperparameters, see `qrnn::qrnn.fit`
#'
#' @inheritParams s_CART
#' @param n.hidden Integer: Number of hidden nodes.
#' @param tau Numeric: tau-quantile.
#' @param n.ensemble Integer: Number of NNs to train.
#' @param iter.max Integer: Max N of iteration of the optimization algorithm.
#' @param n.trials Integer: N of trials. Used to avoid local minima.
#' @param bag Logical: If TRUE, use bagging.
#' @param lower Numeric: Left censoring point.
#' @param eps.seq Numeric: sequence of eps values for the finite smoothing algorithm.
#' @param Th Function: hidden layer transfer function; use `qrnn::sigmoid`, `qrnn::elu`,
#' or `qrnn::softplus` for a nonlinear model and `qrnn::linear` for a linear model.
#' @param Th.prime Function: derivative of hidden layer transfer function.
#' @param penalty Numeric: weight penalty for weight decay regularization.
#' @param ... Additional arguments to be passed to `qrnn::qrnn.fit`.
#'
#' @author E.D. Gennatas
#' @seealso [train_cv] for external cross-validation
#' @family Supervised Learning
#' @export

s_QRNN <- function(x, y = NULL,
                   x.test = NULL, y.test = NULL,
                   x.name = NULL, y.name = NULL,
                   n.hidden = 1,
                   tau = .5,
                   n.ensemble = 5,
                   iter.max = 5000,
                   n.trials = 5,
                   bag = TRUE,
                   lower = -Inf,
                   eps.seq = 2^(-8:-32),
                   Th = qrnn::sigmoid,
                   Th.prime = qrnn::sigmoid.prime,
                   penalty = 0,
                   print.plot = FALSE,
                   plot.fitted = NULL,
                   plot.predicted = NULL,
                   plot.theme = rtTheme,
                   question = NULL,
                   verbose = TRUE,
                   outdir = NULL,
                   save.mod = ifelse(!is.null(outdir), TRUE, FALSE), ...) {
  # Intro ----
  if (missing(x)) {
    print(args(s_QRNN))
    return(invisible(9))
  }
  if (!is.null(outdir)) outdir <- normalizePath(outdir, mustWork = FALSE)
  logFile <- if (!is.null(outdir)) {
    paste0(outdir, "/", sys.calls()[[1]][[1]], ".", format(Sys.time(), "%Y%m%d.%H%M%S"), ".log")
  } else {
    NULL
  }
  start.time <- intro(verbose = verbose, logFile = logFile)
  mod.name <- "QRNN"

  # Dependencies ----
  dependency_check("qrnn")

  # Arguments ----
  if (is.null(y) && NCOL(x) < 2) {
    print(args(s_QRNN))
    stop("y is missing")
  }
  if (is.null(x.name)) x.name <- getName(x, "x")
  if (is.null(y.name)) y.name <- getName(y, "y")
  prefix <- paste0(y.name, "~", x.name)
  if (!verbose) print.plot <- FALSE
  verbose <- verbose | !is.null(logFile)
  if (print.plot) {
    if (is.null(plot.fitted)) plot.fitted <- if (is.null(y.test)) TRUE else FALSE
    if (is.null(plot.predicted)) plot.predicted <- if (!is.null(y.test)) TRUE else FALSE
  } else {
    plot.fitted <- plot.predicted <- FALSE
  }
  if (save.mod && is.null(outdir)) outdir <- paste0("./s.", mod.name)
  if (!is.null(outdir)) outdir <- paste0(normalizePath(outdir, mustWork = FALSE), "/")

  # Data ----
  dt <- prepare_data(x, y, x.test, y.test)
  x <- dt$x
  y <- dt$y
  x.test <- dt$x.test
  y.test <- dt$y.test
  xnames <- dt$xnames
  type <- dt$type
  checkType(type, "Regression", mod.name)
  if (verbose) dataSummary(x, y, x.test, y.test, type)

  # QRNN ----
  if (verbose) msg2("Training QRNN mmodel...", newline.pre = TRUE)
  mod <- qrnn::qrnn.fit(
    x = as.matrix(x), y = as.matrix(y),
    n.hidden = n.hidden,
    tau = tau,
    n.ensemble = n.ensemble,
    iter.max = iter.max,
    n.trials = n.trials,
    bag = bag,
    lower = lower,
    eps.seq = eps.seq,
    Th = Th,
    Th.prime = Th.prime,
    penalty = penalty,
    trace = verbose, ...
  )

  # Fitted ----
  fitted <- rowMeans(qrnn::qrnn.predict(as.matrix(x), mod))
  error.train <- mod_error(y, fitted)
  if (verbose) errorSummary(error.train, mod.name)

  # Predicted ----
  predicted <- error.test <- NULL
  if (!is.null(x.test)) {
    predicted <- rowMeans(qrnn::qrnn.predict(as.matrix(x.test), mod))
    if (!is.null(y.test)) {
      error.test <- mod_error(y.test, predicted)
      if (verbose) errorSummary(error.test, mod.name)
    }
  }

  # Outro ----
  rt <- rtModSet(
    rtclass = "rtMod",
    mod = mod,
    mod.name = mod.name,
    type = type,
    y.train = y,
    y.test = y.test,
    x.name = x.name,
    y.name = y.name,
    xnames = xnames,
    fitted = fitted,
    se.fit = NULL,
    error.train = error.train,
    predicted = predicted,
    se.prediction = NULL,
    error.test = error.test, list,
    question = question
  )

  rtMod.out(
    rt,
    print.plot,
    plot.fitted,
    plot.predicted,
    y.test,
    mod.name,
    outdir,
    save.mod,
    verbose,
    plot.theme
  )

  outro(start.time, verbose = verbose, sinkOff = ifelse(is.null(logFile), FALSE, TRUE))
  rt
} # rtemis::s_QRNN