I think I am doing something wrong with my classifications - r

Question

I would like to see what is wrong with my code when I use the classification methods. My accuracy is very high, and one is 1. I think I did something wrong with my code. Could you please revise the code and tell me code if it is wrong? Since I am very confused about it. I want to predict wine type and wine quality.
The dataset is from -> http://archive.ics.uci.edu/ml/datasets/Wine+Quality
My code:
library(party)
library(tidyverse)
library(RCurl)
library(psych)
library(ggplot2)
library(GGally)
library(mlbench)
library(e1071)
library(caret)
library(rpart)
library(dplyr)
redwine_df<-read.csv("winequality-red.csv")
whitewine_df<-read.csv("winequality-white.csv")
#add categorical values to both sets
redwine_df['wine_type'] <- 'red_wine'
whitewine_df['wine_type'] <- 'white_wine'
is.data.frame(redwine_df)
is.data.frame(whitewine_df)
#merge sets of red wine and white wine
wine <- rbind(redwine_df, whitewine_df)
#change to tibble
wine_tibble<-as_tibble(wine)
wine_tibble
#check the columns
names(wine_tibble)
#dimension
dim(wine_tibble)
#summary
length(which(wine_tibble==0))#just the column citric.acid has 0
summary(wine_tibble)
#drop duplicated values of sets
colSums(is.na(wine_tibble))#is there any na values
summary(duplicated(wine_tibble))#is there any duplicated values
wine_clean <- unique(wine_tibble)
summary(duplicated(wine_clean))
dim(wine_clean)
#Prediction
#Data Preparations - Training and Test Data
w1<-wine_clean %>% mutate(quality_rank =
case_when(quality <= 5 ~ "Poor",
quality == 6 ~ "Normal",
quality >= 7~ "Excellent"))
set.seed(2000)
w1$quality_rank <-as.factor(w1$quality_rank)
#Predict the Wine Quality
inTrain <- createDataPartition(y = w1$quality_rank, p = .8, list = FALSE)
quality_train <- w1 %>% slice(inTrain)
quality_test <- w1 %>% slice(-inTrain)
quality_index <- createFolds(quality_train$quality_rank, k = 10)
quality_train
#1. 1. Conditional Inference Tree (Decision Tree)
install.packages("party")
library(party)
ctreeFit <- quality_train %>% train(quality_rank ~ .,
method = "ctree",
data = .,
tuneLength = 5,
trControl = trainControl(method = "cv", indexOut =
quality_index))
ctreeFit
plot(ctreeFit$finalModel)
#2.Linear Support Vector Machine
svmFit<- svm(quality_rank ~ .,
data = trainset,
type = "C-classification",
kernel = "linear",
scale = FALSE)
svmFit
svmFit$finalModel
#3. C 4.5 Decision Tree
install.packages("RWeka")
library(RWeka)
C45Fit <- quality_train %>% train(quality_rank ~ .,
method = "J48",
data = .,
tuneLength = 5,
trControl = trainControl(method = "cv", indexOut =
quality_index))
C45Fit
C45Fit$finalModel
#4. K-Nearest Neighbors
knnFit <- quality_train %>% train(quality_rank ~ .,
method = "knn",
data = .,
preProcess = "scale",
tuneLength = 5,
tuneGrid=data.frame(k = 1:10),
trControl = trainControl(method = "cv", indexOut =
quality_index))
knnFit
knnFit$finalModel
#5. Naïve Bayes Classifiers
install.packages("klaR")
library(klaR)
NBayesFit <- quality_train %>% train(quality_rank ~ .,
method = "nb",
data = .,
tuneLength = 5,
trControl = trainControl(method = "cv", indexOut =
quality_index))
NBayesFit
#Compare the models for wine quality
resamps <- resamples(list(
ctree = ctreeFit,
C45 = C45Fit,
KNN = knnFit,
NBayes = NBayesFit,
SVM = svmFit))
resamps
#Applying the Chosen Model to the Test Data
summary(resamps)
library(lattice)
bwplot(resamps, layout = c(3, 1))
pr <- predict(knnFit, quality_train)
pr
confusionMatrix(pr, reference = quality_train$quality_rank)
#Predict Wine Type
w2<-wine_clean
w2$wine_type <-as.factor(w2$wine_type)
type_inTrain <- createDataPartition(y = w2$wine_type, p = .9, list = FALSE)
type_train <- w2 %>% slice(type_inTrain)
type_test <- w2 %>% slice(-type_inTrain)
type_index <- createFolds(type_train$wine_type, k = 10)
type_train
#1. Conditional Inference Tree (Decision Tree)
ctreeFit2 <- type_train %>% train(wine_type ~ .,
method = "ctree",
data = .,
tuneLength = 5,
trControl = trainControl(method = "cv", indexOut =
type_index))
ctreeFit2
plot(ctreeFit2$finalModel)
#2.Linear Support Vector Machine
svmFit2 <- type_train %>% train(wine_type ~.,
method = "svmLinear",
data = .,
tuneLength = 5,
trControl = trainControl(method = "cv", indexOut =
type_index))
svmFit2
svmFit2$finalModel
#3. C 4.5 Decision Tree
C45Fit2 <- type_train %>% train(wine_type ~ .,
method = "J48",
data = .,
tuneLength = 5,
trControl = trainControl(method = "cv", indexOut =
type_index))
C45Fit2
C45Fit2$finalModel
#4. K-Nearest Neighbors
knnFit2 <- type_train %>% train(wine_type ~ .,
method = "knn",
data = .,
preProcess = "scale",
tuneLength = 5,
tuneGrid=data.frame(k = 1:10),
trControl = trainControl(method = "cv", indexOut =
type_index))
knnFit2
knnFit2$finalModel
#5. Naïve Bayes Classifiers
NBayesFit2 <- type_train %>% train(wine_type ~ .,
method = "nb",
data = .,
tuneLength = 5,
trControl = trainControl(method = "cv", indexOut =
type_index))
NBayesFit2
#Compare the models for white wine
resamp <- resamples(list(
ctree2 = ctreeFit2,
C452 = C45Fit2,
KNN2 = knnFit2,
NBayes2 = NBayesFit2,
SVM2 = svmFit2
))
resamp
#Applying the Chosen Model to the Test Data
summary(resamp)
library(lattice)
bwplot(resamp, layout = c(3, 1))
pr2 <- predict(svmFit2, type_train)
pr2
confusionMatrix(pr2, reference = type_train$wine_type)
I have tried to changed the train/test spilt to 50%/50?, I changed the set.seed to 123 instead of 2000, but the answer is still the same.