R_dev/rf_example.R

In seb09/rocsy: Summary of ROC curves from a cross validation

set.seed(1909)
data(PimaIndiansDiabetes)

train_index <- caret::createDataPartition(PimaIndiansDiabetes$diabetes, p = 0.75, list = FALSE)

data_train <- PimaIndiansDiabetes %>%
  tibble::as_tibble() %>%
  dplyr::slice(train_index)

data_test <- PimaIndiansDiabetes %>%
  tibble::as_tibble() %>%
  dplyr::slice(-train_index)

fit_ctrl <- caret::trainControl(
  method = "repeatedcv",
  number = 10,
  repeats = 3,
  classProbs = TRUE,
  savePredictions = "all",
  summaryFunction = twoClassSummary
)


fit_train_rf <- caret::train(
  diabetes ~ .,
  data = data_train,
  method = "rf",
  metric="ROC",
  trControl = fit_ctrl
)



fit_train_rf

rocsy(fit_train_rf, pos = "pos", fsum = mean)

diff()

predict(fit_train_rf, newdata = data_test, type="prob") %>%
  dplyr::mutate(obs=pull(data_test, diabetes), pred=factor(ifelse(neg>pos,"neg","pos"))) %>%
  twoClassSummary(lev=c("neg","pos"))