R/s_SVM.R

In egenn/rtemis: Machine Learning and Visualization

# s_SVM.R
# ::rtemis::
# 2016-8 E.D. Gennatas www.lambdamd.org

#' Support Vector Machines (C, R)
#'
#' Train an SVM learner using `e1071::svm`
#'
#' \[gS\] denotes parameters that will be tuned by cross-validation if more than one value is passed.
#' Regarding SVM tuning, the following guide from the LIBSVM authors can be useful:
#' http://www.csie.ntu.edu.tw/~cjlin/papers/guide/guide.pdf
#' They suggest searching for cost = 2 ^ seq(-5, 15, 2) and
#' gamma = 2 ^ seq(-15, 3, 2)
#'
#' @inheritParams s_GLM
#' @inheritParams s_CART
#' @inheritParams resample
#' @param class.weights Float, length = n levels of outcome: Weights for each 
#' outcome class.For classification, `class.weights` takes precedence over 
#' `ifw`, therefore set `class.weights = NULL` if using `ifw`.
#' @param kernel Character: "linear", "polynomial", "radial", "sigmoid"
#' @param degree \[gS\] Integer: Degree for `kernel = "polynomial"`.
#' @param cost \[gS\] Float: Cost of constraints violation; the C constant of the 
#' regularization term in the Lagrange
#'   formulation.
#' @param probability Logical: If TRUE, model allows probability estimates
#' @param gamma \[gS\] Float: Parameter used in all kernels except `linear`
#' @param coef0 \[gS\] Float: Parameter used by kernels `polynomial` and 
#' `sigmoid`
#' @param ... Additional arguments to be passed to `e1071::svm`
#' 
#' @author E.D. Gennatas
#' @seealso [train_cv] for external cross-validation
#' @family Supervised Learning
#' @export

s_SVM <- function(x, y = NULL,
                  x.test = NULL, y.test = NULL,
                  x.name = NULL, y.name = NULL,
                  grid.resample.params = setup.resample("kfold", 5),
                  gridsearch.type = c("exhaustive", "randomized"),
                  gridsearch.randomized.p = .1,
                  class.weights = NULL,
                  ifw = TRUE,
                  ifw.type = 2,
                  upsample = FALSE,
                  downsample = FALSE,
                  resample.seed = NULL,
                  kernel = "radial",
                  degree = 3,
                  gamma = NULL,
                  coef0 = 0,
                  cost = 1,
                  probability = TRUE,
                  metric = NULL,
                  maximize = NULL,
                  plot.fitted = NULL,
                  plot.predicted = NULL,
                  print.plot = FALSE,
                  plot.theme = rtTheme,
                  n.cores = rtCores,
                  question = NULL,
                  verbose = TRUE,
                  grid.verbose = verbose,
                  outdir = NULL,
                  save.mod = ifelse(!is.null(outdir), TRUE, FALSE), ...) {

  # Intro ----
  if (missing(x)) {
    print(args(s_SVM))
    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 <- "SVM"

  # Dependencies ----
  dependency_check("e1071")

  # Arguments ----
  if (is.null(y) & NCOL(x) < 2) {
    print(args(s_SVM))
    stop("y is missing")
  }
  if (is.null(x.name)) x.name <- getName(x, "x")
  if (is.null(y.name)) y.name <- getName(y, "y")
  if (!verbose) print.plot <- FALSE
  verbose <- verbose | !is.null(logFile)
  if (save.mod & is.null(outdir)) outdir <- paste0("./s.", mod.name)
  if (!is.null(outdir)) outdir <- paste0(normalizePath(outdir, mustWork = FALSE), "/")
  gridsearch.type <- match.arg(gridsearch.type)

  # Data ----
  dt <- prepare_data(x, y, x.test, y.test,
                    ifw = ifw,
                    ifw.type = ifw.type,
                    upsample = upsample,
                    downsample = downsample,
                    resample.seed = resample.seed,
                    verbose = verbose)
  x <- data.matrix(dt$x)
  y <- dt$y
  x.test <- dt$x.test
  if (!is.null(x.test)) x.test <- data.matrix(x.test)
  y.test <- dt$y.test
  xnames <- dt$xnames
  type <- dt$type
  checkType(type, c("Classification", "Regression"), mod.name)
  .class.weights <- if (is.null(class.weights) & ifw) dt$class.weights else class.weights
  x0 <- if (upsample | downsample) dt$x0 else x
  y0 <- if (upsample | downsample) dt$y0 else y
  if (verbose) dataSummary(x, y, x.test, y.test, type)
  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 (is.null(gamma)) gamma <- 1 / NCOL(x)
  if (type == "Classification") nlevels <- length(levels(y))

  # Grid Search ----
  if (is.null(metric)) {
    if (type == "Classification") {
      metric <- "Balanced Accuracy"
      if (is.null(maximize)) maximize <- TRUE
    } else if (type == "Regression") {
      metric <- "MSE"
      if (is.null(maximize)) maximize <- FALSE
    }
  }

  if (is.null(maximize)) {
    maximize <- if (type == "Classification") TRUE else FALSE
  }

  gs <- NULL
  if (kernel == "linear") {
    if (gridCheck(cost)) {
      gs <- gridSearchLearn(x0, y0, mod.name,
                            resample.params = grid.resample.params,
                            grid.params = list(cost = cost),
                            fixed.params = list(kernel = "linear",
                                                class.weights = class.weights,
                                                ifw = ifw,
                                                ifw.type = ifw.type,
                                                upsample = upsample,
                                                downsample = downsample,
                                                resample.seed = resample.seed),
                            search.type = gridsearch.type,
                            randomized.p = gridsearch.randomized.p,
                            metric = metric,
                            maximize = maximize,
                            verbose = verbose, grid.verbose = grid.verbose, n.cores = n.cores)
      cost <- gs$best.tune$cost
    }
  } else if (kernel == "polynomial") {
    if (gridCheck(cost, degree, gamma, coef0)) {
      gs <- gridSearchLearn(x, y, mod.name,
                            resample.params = grid.resample.params,
                            grid.params = list(cost = cost, degree = degree, gamma = gamma, coef0 = coef0),
                            fixed.params = list(kernel = "polynomial",
                                                class.weights = class.weights,
                                                ifw = ifw,
                                                ifw.type = ifw.type,
                                                upsample = upsample,
                                                downsample = downsample,
                                                resample.seed = resample.seed),
                            metric = metric,
                            maximize = maximize,
                            verbose = verbose, grid.verbose = grid.verbose, n.cores = n.cores)
      cost <- gs$best.tune$cost
      degree <- gs$best.tune$degree
      gamma <- gs$best.tune$gamma
      coef0 <- gs$best.tune$coef0
    }
  } else if (kernel == "sigmoid") {
    if (gridCheck(cost, gamma, coef0)) {
      gs <- gridSearchLearn(x, y, mod.name,
                            resample.params = grid.resample.params,
                            grid.params = list(cost = cost, gamma = gamma, coef0 = coef0),
                            fixed.params = list(kernel = "sigmoid",
                                                class.weights = class.weights,
                                                ifw = ifw,
                                                ifw.type = ifw.type,
                                                upsample = upsample,
                                                downsample = downsample,
                                                resample.seed = resample.seed),
                            metric = metric,
                            maximize = maximize,
                            verbose = verbose, grid.verbose = grid.verbose, n.cores = n.cores)
      cost <- gs$best.tune$cost
      gamma <- gs$best.tune$gamma
      coef0 <- gs$best.tune$coef0
    }
  } else {
    if (gridCheck(cost, gamma)) {
      gs <- gridSearchLearn(x, y, mod.name,
                            resample.params = grid.resample.params,
                            grid.params = list(cost = cost, gamma = gamma),
                            fixed.params = list(kernel = "radial",
                                                class.weights = class.weights,
                                                ifw = ifw,
                                                ifw.type = ifw.type,
                                                upsample = upsample,
                                                downsample = downsample,
                                                resample.seed = resample.seed),
                            metric = metric,
                            maximize = maximize,
                            verbose = verbose, grid.verbose = grid.verbose, n.cores = n.cores)
      cost <- gs$best.tune$cost
      gamma <- gs$best.tune$gamma
    }
  }

  if (kernel == "linear") {
    parameters <- list(kernel = kernel, cost = cost, class.weights = .class.weights)
  } else if (kernel == "polynomial") {
    parameters <- list(kernel = kernel, cost = cost, gamma = gamma, coef0 = coef0, degree = degree, class.weights = .class.weights)
  } else if (kernel == "sigmoid") {
    parameters <- list(kernel = kernel, cost = cost, gamma = gamma, coef0 = coef0, class.weights = .class.weights)
  } else {
    parameters <- list(kernel = kernel, cost = cost, gamma = gamma, class.weights = .class.weights)
  }

  # e1071::svm ----
  if (verbose) msg2("Training SVM", type, "with", kernel, "kernel...", newline.pre = TRUE)
  mod <- e1071::svm(x = x, y = y,
                    kernel = kernel,
                    degree = degree,
                    cost = cost,
                    gamma = gamma,
                    coef0 = coef0,
                    probability = probability,
                    class.weights = .class.weights, ...)

  # Fitted ----
  fitted.prob <- predict(mod, x, probability = TRUE)
  fitted.prob <- attr(fitted.prob, "probabilities")[, levels(y)[1]]
  fitted <- predict(mod)

  if (type == "Regression") fitted <- as.numeric(fitted)
  error.train <- mod_error(y, fitted, fitted.prob)
  if (verbose) errorSummary(error.train, mod.name)

  # Predicted ----
  predicted.prob <- predicted <- error.test <- NULL
  if (!is.null(x.test)) {
    predicted.prob <- predict(mod, x.test, probability = TRUE)
    predicted.prob <- attr(predicted.prob, "probabilities")[, levels(y)[1]]
    predicted <- predict(mod, x.test)
    if (!is.null(y.test)) {
      if (type == "Regression") predicted <- as.numeric(predicted)
      error.test <- mod_error(y.test, predicted, predicted.prob)
      if (verbose) errorSummary(error.test, mod.name)
    }
  }

  # Outro ----
  rt <- rtModSet(rtclass = "rtMod",
                 mod = mod,
                 mod.name = mod.name,
                 type = type,
                 gridsearch = gs,
                 parameters = parameters,
                 call = call,
                 y.train = y,
                 y.test = y.test,
                 x.name = x.name,
                 y.name = y.name,
                 xnames = xnames,
                 fitted = fitted,
                 fitted.prob = fitted.prob,
                 se.fit = NULL,
                 error.train = error.train,
                 predicted = predicted,
                 predicted.prob = predicted.prob,
                 se.prediction = NULL,
                 error.test = error.test,
                 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_SVM