R studio plotting issue - "plot.new has not been called yet"

[mshyu]

Hi all, I'm new to R language. I'm trying to predict which variable relates to the SAH patients' death. I got several errors. How should I fix it? Thanks in advance!!
Here's the code:

library(caret)
library(pROC)
fitControl <- trainControl(method = "repeatedcv",
number = 10,
repeats= 10,
classProbs = TRUE,
summaryFunction = twoClassSummary()) # Estimate class probabilities

data<-read.csv("full_train_binary.csv")

data_training<-subset(data[which(data$X==0),])
data_testing<-subset(data[which(data$X==1),])

training<-data_training
testing<-data_testing

cols_remove <- c("patient_sk","New_admitted_dt_tm", "New_discharge_dt_tm")

training<-training[,!(colnames(training)%in%cols_remove)]
testing<-testing[,!(colnames(testing)%in%cols_remove)]

set.seed(825)

start.time <- Sys.time()

svm_one <- train(death~., data = training,
method = 'svmRadial',
trControl = fitControl,
verbose = FALSE,
tunelength=5,
metric="ROC")
svm_one

end.time <- Sys.time()
time.taken <- end.time - start.time
time.taken

svm_one_pred <- predict(svm_one, newdata=testing,type = 'prob')
roc_svm_one <- roc(testing$death, as.vector(svm_one_pred[,1]))
pROC::auc(roc_svm_one)

Here's the result:

library(caret)
library(pROC)
fitControl <- trainControl(method = "repeatedcv",
                              number = 10, 
                              repeats= 10,
                              classProbs =  TRUE,
                              summaryFunction = twoClassSummary())            # Estimate class probabilities
Error in levels(data[, "pred"]) : 
  argument "data" is missing, with no default

data<-read.csv("/Users/mding/Downloads/1976/R/full_train_binary.csv")

data_training<-subset(data[which(data$X==0),])
data_testing<-subset(data[which(data$X==1),])
training<-data_training
testing<-data_testing

cols_remove <- c("patient_sk","New_admitted_dt_tm", "New_discharge_dt_tm")

training<-training[,!(colnames(training)%in%cols_remove)]
testing<-testing[,!(colnames(testing)%in%cols_remove)]

set.seed(825)

start.time <- Sys.time()

svm_one <- train(death~., data = training, 
                  method = 'svmRadial',                     # model 1: svm with radial kernel 
                  trControl = fitControl, 
                  verbose = FALSE,
                  tunelength=5,
                  metric="ROC")
Error: Every row has at least one missing value were found
svm_one
Support Vector Machines with Radial Basis Function Kernel 

4911 samples
1954 predictors
   2 classes: 'False', 'True' 

No pre-processing
Resampling: Cross-Validated (5 fold) 
Summary of sample sizes: 3928, 3928, 3929, 3930, 3929 
Resampling results across tuning parameters:

  sigma         C          Accuracy   Kappa        
  1.976927e-05  192.56972  0.7448586  -0.0004065338
  2.778991e-05  242.26352  0.7446545   0.0007460142
  3.273858e-05   14.39494  0.7450623   0.0000000000

Accuracy was used to select the optimal model using the largest value.
The final values used for the model were sigma = 3.273858e-05 and C
 = 14.39494.

end.time <- Sys.time()
time.taken <- end.time - start.time
time.taken
Time difference of 0.395869 secs
 
 
svm_one_pred <- predict(svm_one, newdata=testing,type = 'prob')
Error in eval(predvars, data, env) : object 'patient_sk' not found
roc_svm_one <- roc(testing$death, as.vector(svm_one_pred[,1]))
Error in as.vector(svm_one_pred[, 1]) : object 'svm_one_pred' not found
pROC::auc(roc_svm_one)
Error in pROC::auc(roc_svm_one) : object 'roc_svm_one' not found

[technocrat]

Cockroaches seldom travel alone. Me

The last error message is Error in pROC::auc(roc_svm_one) : object 'roc_svm_one' not found, but it's proceeded by others. So, first things first:

Error in levels(data[, "pred"]) : 
  argument "data" is missing, with no default

The fix is to create the data object before setting fitControl. That may or may not do it. But it's the place to start.