library(caret)
library(C50)
wine <- read.csv('wine2.csv',header = T,stringsAsFactors = T,fileEncoding = 'euc-kr')
head(wine)
prop.table(table(wine$Type))
colSums(is.na(wine))
set.seed(1)
train_num <- createDataPartition(wine$Type,p=0.9,list=F)
train_data <- wine[train_num,]
test_data <- wine[-train_num,]
options(scipen=999)
for(i in 1:100){
wine_model <- C5.0(train_data[,-1],train_data[,1],trials = i)
result_train <- predict(wine_model,train_data[,-1])
result_test <- predict(wine_model,test_data[,-1])
tr <- sum(result_train == train_data[,1])/length(result_train) * 100
te <- sum(result_test == test_data[,1])/length(result_test) * 100
print(paste('trials = ',i,' 훈련 정확도: ',tr,' 테스트 정확도: ',te))
}
trials 3 이후부터는 훈련,테스트 모두 100