R/TaskSurv_zzz.R

In mlr-org/mlr3proba: Probabilistic Supervised Learning for 'mlr3'

#' @title German Breast Cancer Study Survival Task
#'
#' @name mlr_tasks_gbsg
#' @templateVar type Surv
#' @templateVar task_type survival
#' @templateVar id gbsg
#' @templateVar data gbsg
#' @template task
#' @template seealso_task
#'
#' @section Pre-processing:
#' - Removed column `pid`.
#' - Column `meno` has been converted to `factor` and 0/1 values have been
#' replaced with `premenopausal` and `postmenopausal` respectively.
#' - Column `hormon` has been converted to `factor` and 0/1 values have been
#' replaced with `no` and `yes` respectively.
#' - Column `grade` has been converted to `factor`.
#' - Renamed target column `rfstime` to `time`.
NULL

load_gbsg = function() {
  data = survival::gbsg
  data$pid = NULL
  data$meno = factor(ifelse(data$meno == 0, "premenopausal", "postmenopausal"),
                     levels = c("premenopausal", "postmenopausal"))
  data$hormon = factor(ifelse(data$hormon == 0, "no", "yes"),
                       levels = c("no", "yes"))
  data$grade = factor(data$grade)
  colnames(data)[colnames(data) == "rfstime"] = "time"

  b = as_data_backend(data)
  task = TaskSurv$new("gbsg", b, time = "time", event = "status",
                      label = "German Breast Cancer")
  b$hash = task$man = "mlr3proba::mlr_tasks_gbsc"

  task
}

#' @title Primary Biliary Cholangitis Survival Task
#'
#' @name mlr_tasks_pbc
#' @templateVar type Surv
#' @templateVar task_type survival
#' @templateVar id pbc
#' @templateVar data pbc
#' @template task
#' @template seealso_task
#'
#' @section Pre-processing:
#' - Removed column `id`.
#' - Kept only complete cases (no missing values).
#' - Column `age` has been converted to `integer`.
#' - Columns `trt`, `stage`, `hepato`, `edema` and `ascites` have been converted
#' to `factor`s.
#' - Column `trt` has levels `Dpenicillmain` and `placebo` instead of 1 and 2.
#' - Column `status` has 1 for death and 0 for censored or transplant.
NULL

load_pbc = function() {
  data = survival::pbc
  data = stats::na.omit(data)
  data$id = NULL
  data = map_at(data, c("age"), as.integer)
  data = map_at(data, c("spiders", "hepato", "edema", "ascites"), as.factor)
  data$trt = factor(ifelse(data$trt == 1, "Dpenicillmain", "placebo"),
                    levels = c("Dpenicillmain", "placebo"))
  data$stage = factor(data$stage)
  data$status[data$status > 0] = data$status[data$status > 0] - 1

  b = as_data_backend(data)
  task = TaskSurv$new("pbc", b, time = "time", event = "status",
                      label = "Primary Biliary Cholangitis")
  b$hash = task$man = "mlr3proba::mlr_tasks_pbc"

  task
}

#' @title Monoclonal Gammopathy Survival Task
#'
#' @name mlr_tasks_mgus
#' @templateVar type Surv
#' @templateVar task_type survival
#' @templateVar id mgus
#' @templateVar data mgus
#' @template task
#' @template seealso_task
#'
#' @section Pre-processing:
#' - Removed columns `id`, `pcdx` and `pctime`.
#' - Renamed target columns from (`fultime`, `death`) to (`time`, `status`).
#' - Kept only complete cases (no missing values).
NULL

load_mgus = function() {
  data = survival::mgus
  data[, c("id", "pcdx", "pctime")] = NULL
  colnames(data)[colnames(data) == "futime"] = "time"
  colnames(data)[colnames(data) == "death"] = "status"
  data = stats::na.omit(data)

  b = as_data_backend(data)
  task = TaskSurv$new("mgus", b, time = "time", event = "status", label = "MGUS")
  b$hash = task$man = "mlr3proba::mlr_tasks_mgus"

  task
}

#' @title Veteran Survival Task
#'
#' @name mlr_tasks_veteran
#' @templateVar type Surv
#' @templateVar task_type survival
#' @templateVar id veteran
#' @templateVar data veteran
#' @template task
#' @template seealso_task
#'
#' @section Pre-processing:
#' - Columns `age`, `time`, `status`, `diagtime` and `karno` have been converted
#' to `integer`.
#' - Columns `trt`, `prior` have been converted to `factor`s. Prior therapy
#' values are `no`/`yes` instead of 0/10.
NULL

load_veteran = function() {
  data = survival::veteran
  data = map_at(data, c("age", "time", "status", "diagtime", "karno"), as.integer)
  data = map_at(data, c("trt", "prior"), as.factor)
  data$trt = factor(data$trt, levels = c("1", "2"))
  data$prior = factor(ifelse(data$prior == 0, "no", "yes"), levels = c("no", "yes"))

  b = as_data_backend(data)
  task = TaskSurv$new("veteran", b, time = "time", event = "status", label = "Veteran")
  b$hash = task$man = "mlr3proba::mlr_tasks_veteran"

  task
}

#' @title Rats Survival Task
#'
#' @name mlr_tasks_rats
#' @templateVar type Surv
#' @templateVar task_type survival
#' @templateVar id rats
#' @templateVar data rats
#' @template task
#' @template seealso_task
#'
#' @section Pre-processing:
#' - Column `sex` has been converted to a `factor`, all others have been
#' converted to `integer`.
NULL

load_rats = function() {
  data = survival::rats
  data = map_at(data, c("rx", "time", "status"), as.integer)
  data$sex = factor(data$sex, levels = c("f", "m"))

  b = as_data_backend(data)
  task = TaskSurv$new("rats", b, time = "time", event = "status", label = "Rats")
  b$hash = task$man = "mlr3proba::mlr_tasks_rats"

  task
}

#' @title Lung Cancer Survival Task
#'
#' @name mlr_tasks_lung
#'
#' @templateVar type Surv
#' @templateVar task_type survival
#' @templateVar id lung
#' @templateVar data lung
#' @template task
#' @template seealso_task
#'
#' @section Pre-processing:
#' - Column `inst` has been removed.
#' - Column `sex` has been converted to a `factor`, all others have been
#' converted to `integer`.
#' - Kept only complete cases (no missing values).
NULL

load_lung = function() {
  data = survival::lung
  data$inst = NULL
  data = map_dtc(data, as.integer)
  data$status = as.integer(data$status == 2L)
  data$sex = factor(ifelse(data$sex == 1L, "m", "f"), levels = c("f", "m"))
  data = stats::na.omit(data)

  b = as_data_backend(data)
  task = TaskSurv$new("lung", b, time = "time", event = "status", label = "Lung Cancer")
  b$hash = task$man = "mlr3proba::mlr_tasks_lung"

  task
}

#' @title ACTG 320 Survival Task
#'
#' @name mlr_tasks_actg
#'
#' @templateVar type Surv
#' @templateVar task_type survival
#' @templateVar id actg
#' @templateVar data actg
#' @template task
#' @template seealso_task
#'
#' @section Pre-processing:
#' - Column `sex` has been renamed to `sexF` and `censor` has been renamed to `status`.
#' - Columns `id`, `time_d`, and `censor_d` have been removed so target is `time`
#' to AIDS diagnosis (in days).
NULL

load_actg = function() {
  data = load_dataset("actg", "mlr3proba")
  data[, c("id", "time_d", "censor_d")] = NULL
  colnames(data)[6L] = "sexF"
  colnames(data)[2L] = "status"

  b = as_data_backend(data)
  task = TaskSurv$new("actg", b, time = "time", event = "status", label = "ACTG 320")
  b$hash = task$man = "mlr3proba::mlr_tasks_actg"

  task
}

#' @title German Breast Cancer Study Survival Task
#'
#' @name mlr_tasks_gbcs
#'
#' @templateVar type Surv
#' @templateVar task_type survival
#' @templateVar id gbcs
#' @templateVar data gbcs
#' @template task
#' @template seealso_task
#'
#' @section Preprocessing:
#' - Column `id` and all date columns have been removed, as well as `rectime`
#' and `censrec`.
#' - Target columns (`survtime`, `censdead`) have been renamed to (`time`, `status`).
NULL

load_gbcs = function() {
  data = load_dataset("gbcs", "mlr3proba")
  data[, c("id", "diagdate", "recdate", "deathdate", "rectime", "censrec")] = NULL
  colnames(data)[9:10] = c("time", "status")

  b = as_data_backend(data)
  task = TaskSurv$new("gbcs", b, time = "time", event = "status", label = "German Breast Cancer")
  b$hash = task$man = "mlr3proba::mlr_tasks_gbcs"

  task
}

#' @title GRACE 1000 Survival Task
#'
#' @name mlr_tasks_grace
#'
#' @templateVar type Surv
#' @templateVar task_type survival
#' @templateVar id grace
#' @templateVar data grace
#' @template task
#' @template seealso_task
#'
#' @section Preprocessing:
#' - Column `id` is removed.
#' - Target columns (`days`, `death`) have been renamed to (`time`, `status`).
NULL

load_grace = function() {
  data = load_dataset("grace", "mlr3proba")
  data[, "id"] = NULL
  colnames(data)[1:2] = c("time", "status")

  b = as_data_backend(data)
  task = TaskSurv$new("grace", b, time = "time", event = "status", label = "GRACE 1000")
  b$hash = task$man = "mlr3proba::mlr_tasks_grace"

  task
}

#' @title Worcester Heart Attack Study (WHAS) Survival Task
#'
#' @name mlr_tasks_whas
#'
#' @templateVar type Surv
#' @templateVar task_type survival
#' @templateVar id whas
#' @templateVar data whas
#' @template task
#' @template seealso_task
#'
#' @section Preprocessing:
#' - Columns `id`, `yrgrp`, and `dstat` are removed.
#' - Column `sex` is renamed to `sexF`, `lenfol` to `time`, and `fstat` to `status`.
#' - Target is total follow-up time from hospital admission.
NULL

load_whas = function() {
  data = load_dataset("whas", "mlr3proba")
  data[, c("id", "yrgrp", "dstat")] = NULL
  colnames(data)[2L] = "sexF"
  colnames(data)[10:11] = c("time", "status")

  b = as_data_backend(data)
  task = TaskSurv$new("whas", b, time = "time", event = "status",
                      label = "Worcester Heart Attack")
  b$hash = task$man = "mlr3proba::mlr_tasks_whas"

  task
}

register_task("gbsg", load_gbsg)
register_task("pbc", load_pbc)
register_task("mgus", load_mgus)
register_task("veteran", load_veteran)
register_task("rats", load_rats)
register_task("lung", load_lung)
register_task("actg", load_actg)
register_task("gbcs", load_gbcs)
register_task("grace", load_grace)
register_task("whas", load_whas)