gist_3

result_matrix = matrix(nrow = 200, ncol = 3)
for (i in 1:200){
  
  set.seed(i)
  
  n=nrow(df)
  size.train=floor(n*0.50)
  size.valid=floor(n*0.50)
  id.train=sample(1:n,size.train,replace=FALSE)
  id.valid=sample(setdiff(1:n,id.train),size.valid,replace=FALSE)
  mydata.train=df[id.train,]
  mydata.valid=df[id.valid,]
  
  ######## RANDOM FOREST #########
  rf=randomForest(diagnosis~.,data=mydata.train,ntree=250, mtry = 8)
  predrf=predict(rf,newdata=mydata.valid)
  accuracy_forest = mean(predrf==mydata.valid$diagnosis)
  result_matrix[i,1] =accuracy_forest
  
  ######## SUPPORT VECTOR MACHINE #########
  mysvm = svm(diagnosis~., data = mydata.train, kernel="polynomial", cost=5, degree=3)
  pred_svm_optimal = predict(mysvm, mydata.valid)
  accuracy_svm = mean(pred_svm_optimal==mydata.valid$diagnosis)
  result_matrix[i,2] = accuracy_svm
  
  ######## LOGISTIC REGRESSION #########
  logistic = glm(mydata.train$diagnosis~., data = mydata.train, family = binomial)
  pred = round(predict(logistic, type = "response", newdata=mydata.valid))
  accuracy_logistic = mean(pred==mydata.valid$diagnosis)
  result_matrix[i,3] = accuracy_logistic
  
}

accuracy_forest = mean(result_matrix[,1])
accuracy_svm = mean(result_matrix[,2])
accuracy_logistic = mean(result_matrix[,3])