cindyangelira

# Benchamrk----
resample <- rsmp("cv", folds = 5)
models <- benchmark_grid(task, learners, resample)
bnchmrk <- benchmark(models)

# Aggregate result
measures <- msrs(c("surv.cindex", "surv.graf"))
bnchmrk$aggregate(measures)
bm <- fortify(bnchmrk)
aggr <- bnchmrk$aggregate(measures)

cindyangelira

# Create learners----
learners <- lrns(
  paste0("surv.", c("coxtime","deephit", "deepsurv", "loghaz", "pchazard")),
  frac = 0.3, activation = "relu",
  dropout = 0.1, 
  early_stopping = TRUE, 
  epochs = 10,
  batch_size = 32L
  )
learners <- c(learners, lrns(c("surv.kaplan", "surv.coxph")))

cindyangelira

task <- TaskSurv$new("patient_id", backend = df_clean, time = "time", event = "event", type = "right")
task

cindyangelira

# Pre processing----
df <- df %>% select(-6)
colnames(df)[c(4,5,6)] <- c("age", "begin", "disease")

#  Encoding the  categorical feature----
library(caret)
dmy <- dummyVars(" ~ .", data = df)
df_clean <- data.frame(predict(dmy, newdata = df)) %>% select(-c(1,6))
head(df_clean)

cindyangelira

set.seed(1)
df <- read_csv("Dialysis.csv")
head(df)

cindyangelira

required_packages <- c("tidyverse", "survivalmodels", "mlr3benchmark", 
                      "mlr3pipelines", "mlr3proba", "mlr3tuning", "mlr3extralearners")
install.packages(required_packages)

cindyangelira

# Read Dialysis dataset from SurvSet.data

from SurvSet.data import SurvLoader
loader = SurvLoader()
df, ref = loader.load_dataset(ds_name='Dialysis').values()
df.to_csv('Dialysis.csv', index=False)