如何使用 for 循环运行代码 1000 次？

Question

我有这段代码，每次运行时都会产生不同的“auc”。我想运行此代码 1000 次，以计算 1000 个存储的 AUC 的平均值。我如何申请循环来做我正在寻找的东西？如果你们为我做这件事并粘贴新代码，将不胜感激。因为我从 4 天以来一直在尝试解决。

#iris is a built-in dataset
main_df<- iris
# extract data for "setosa"
setosa <-main_df[main_df$Species=="setosa" ,]
# extract data for "virginica"
virginica<-main_df[main_df$Species=="virginica" ,]
#merge "setosa" and "virginica" as new dataset
df<- rbind(setosa,virginica)

Cross.AUC<-rep(0,1000) # create a vector of zeros, here will be stored the auc values from each of 1000 runs
for (i in seq(1:1000)) {
#---------------devide data into two datasets 70:30 train:test ----
#---------------train dataset
#select randomly 70% of setosa, generates a 35-by-5 matrix
setosa_70<-setosa[sample(nrow(setosa),round(0.7*dim(setosa)[1])),]
#select randomly 70% of virginica, generates a 35-by-5 matrix
virginica_70<-virginica[sample(nrow(virginica),round(0.7*dim(virginica)[1])),]
#merge setosa and virginica
train<-rbind(setosa_70,virginica_70)
#convert "setosa" to "0" and "virginica" to 1""
train$Species<-ifelse(train$Species=="setosa",0,1)
#select 1st, 2nd and 5th columns
train <-subset(train,select = c(1,2,5))
#--------------test dataset
#select randomly 30% of setosa, generates a 15-by-5 matrix
setosa_30<-setosa[sample(nrow(setosa),round(0.3*dim(setosa)[1])),]
#select randomly 30% of virginica, generates a 15-by-5 matrix
virginica_30<-virginica[sample(nrow(virginica),round(0.3*dim(virginica)[1])),]
#merge setosa and virginica
test<-rbind(setosa_30,virginica_30)
#convert "setosa" to "0" and "virginica" to 1""
test$Species<-ifelse(test$Species=="setosa",0,1)
#select 1st, 2nd and 5th columns
test <-subset(test,select = c(1,2,5))
#merge "train" and "test"
train_test<-rbind(train,test)
#--Model_1--
model <-glm(Species~., family = binomial(link = "logit"),data = train_test)
# install.packages("ROCR")
library(ROCR)
p <- predict(model, newdata=test, type="response")
pr <- prediction(p, test$Species)
auc <- performance(pr, measure = "auc")
auc <- auc@y.values[[1]]
AUC[i]<-auc
}

Answer 1

将AUC 替换为Cross.AUC。您没有在您可以在for 循环中访问的范围内定义任何名为AUC 的对象。

#iris is a built-in dataset
main_df<- iris
# extract data for "setosa"
setosa <-main_df[main_df$Species=="setosa" ,]
# extract data for "virginica"
virginica<-main_df[main_df$Species=="virginica" ,]
#merge "setosa" and "virginica" as new dataset
df<- rbind(setosa,virginica)

Cross.AUC<-rep(0,1000) # create a vector of zeros, here will be stored the auc values from each of 1000 runs
for (i in seq(1:1000)) {
  #---------------devide data into two datasets 70:30 train:test ----
  #---------------train dataset
  #select randomly 70% of setosa, generates a 35-by-5 matrix
  setosa_70<-setosa[sample(nrow(setosa),round(0.7*dim(setosa)[1])),]
  #select randomly 70% of virginica, generates a 35-by-5 matrix
  virginica_70<-virginica[sample(nrow(virginica),round(0.7*dim(virginica)[1])),]
  #merge setosa and virginica
  train<-rbind(setosa_70,virginica_70)
  #convert "setosa" to "0" and "virginica" to 1""
  train$Species<-ifelse(train$Species=="setosa",0,1)
  #select 1st, 2nd and 5th columns
  train <-subset(train,select = c(1,2,5))
  #--------------test dataset
  #select randomly 30% of setosa, generates a 15-by-5 matrix
  setosa_30<-setosa[sample(nrow(setosa),round(0.3*dim(setosa)[1])),]
  #select randomly 30% of virginica, generates a 15-by-5 matrix
  virginica_30<-virginica[sample(nrow(virginica),round(0.3*dim(virginica)[1])),]
  #merge setosa and virginica
  test<-rbind(setosa_30,virginica_30)
  #convert "setosa" to "0" and "virginica" to 1""
  test$Species<-ifelse(test$Species=="setosa",0,1)
  #select 1st, 2nd and 5th columns
  test <-subset(test,select = c(1,2,5))
  #merge "train" and "test"
  train_test<-rbind(train,test)
  #--Model_1--
  model <-glm(Species~., family = binomial(link = "logit"),data = train_test)
  # install.packages("ROCR")
  library(ROCR)
  p <- predict(model, newdata=test, type="response")
  pr <- prediction(p, test$Species)
  auc <- performance(pr, measure = "auc")
  auc <- auc@y.values[[1]]
  Cross.AUC[i]<-auc
}

cat('Mean AUC:', mean(Cross.AUC), '\n')