RaphaelS1

#' riscitation - Create .ris file from R citation
#'
#' Suggests:
#'  Package `berryFunctions` if `opendelete = TRUE`.
#'
#' Arguments
#'  `pkg:character(1)` - Name of package to cite
#'  `path:character(1)` - Path to write file too, should include file name but not extension
#'  `opendelete:logical(1)` - If `TRUE` (default), opens the .ris file in the default application then deletes
#'                            the file.

RaphaelS1

options(repos=c(
  mlrorg = 'https://mlr-org.r-universe.dev',
  raphaels1 = 'https://raphaels1.r-universe.dev',
  CRAN = 'https://cloud.r-project.org'
))
install.packages(c("ggplot2", "mlr3benchmark", "mlr3pipelines", "mlr3proba", "mlr3tuning", 
                   "survivalmodels", "mlr3extralearners"))

RaphaelS1

library(survivalmodels)

install_pycox(pip = TRUE, install_torch = TRUE)
install_keras(pip = TRUE, install_tensorflow = TRUE)

RaphaelS1

set_seed(1234)

RaphaelS1

library(mlr3)
library(mlr3proba)

## get the `whas` task from mlr3proba
whas <- tsk("whas")

## create our own task from the rats dataset
rats_data <- survival::rats

## convert characters to factors

RaphaelS1

library(paradox)

search_space <- ps(
 ## p_dbl for numeric valued parameters
 dropout = p_dbl(lower = 0, upper = 1),
 weight_decay = p_dbl(lower = 0, upper = 0.5),
 learning_rate = p_dbl(lower = 0, upper = 1),
 
 ## p_int for integer valued parameters
 nodes = p_int(lower = 1, upper = 32),

RaphaelS1

library(mlr3tuning)

create_autotuner <- function(learner) {
 AutoTuner$new(
   learner = learner,
   search_space = search_space,
   resampling = rsmp("holdout"),
   measure = msr("surv.cindex"),
   terminator = trm("evals", n_evals = 2),
   tuner = tnr("random_search")

RaphaelS1

## learners are stored in mlr3extralearners
library(mlr3extralearners)

## load learners
learners <- lrns(
 paste0("surv.", c("coxtime", "deephit", "deepsurv", "loghaz", "pchazard")),
 frac = 0.3, early_stopping = TRUE, epochs = 10, optimizer = "adam"
)
 
# apply our function

RaphaelS1

library(mlr3pipelines)

create_pipeops <- function(learner) {
 po("encode") %>>% po("scale") %>>% po("learner", learner)
}

## apply our function
learners <- lapply(learners, create_pipeops)

RaphaelS1

## select holdout as the resampling strategy
resampling <- rsmp("cv", folds = 3)

## add KM and CPH
learners <- c(learners, lrns(c("surv.kaplan", "surv.coxph")))
design <- benchmark_grid(tasks, learners, resampling)
bm <- benchmark(design)
               
## Aggreggate with Harrell's C and Integrated Graf Score
msrs <- msrs(c("surv.cindex", "surv.graf"))