SVM Proyek Bidang Industri

# Library
library(mlbench)

## Warning: package 'mlbench' was built under R version 4.4.3

library(e1071)

## Warning: package 'e1071' was built under R version 4.4.3

library(class)
library(rpart)

## Warning: package 'rpart' was built under R version 4.4.3

library(caret)

## Warning: package 'caret' was built under R version 4.4.3

## Loading required package: ggplot2

## Warning: package 'ggplot2' was built under R version 4.4.3

## Loading required package: lattice

library(ggplot2)
library(gridExtra)

## Warning: package 'gridExtra' was built under R version 4.4.3

library(dplyr)

## 
## Attaching package: 'dplyr'

## The following object is masked from 'package:gridExtra':
## 
##     combine

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

library(tidyr)

# Simulasi Data Baru
set.seed(123)
n <- 200
df <- data.frame(
  Tekanan = rnorm(n, mean = 120, sd = 10),
  Suhu = rnorm(n, mean = 650, sd = 20),
  Waktu = rnorm(n, mean = 30, sd = 5),
  Kadar_Zat = rnorm(n, mean = 2.5, sd = 0.3),
  Konsentrasi = runif(n, min = 0.6, max = 0.9),
  Ketebalan = rnorm(n, mean = 5.5, sd = 0.2),
  Logam_Legasi = rnorm(n, mean = 8, sd = 0.5),
  Usia_Mesin = sample(3:10, n, replace = TRUE),
  Kualitas = sample(c("OK", "Defect"), n, replace = TRUE, prob = c(0.7, 0.3))
)
df$Kualitas <- as.factor(df$Kualitas)

# Split data
set.seed(42)
index <- createDataPartition(df$Kualitas, p = 0.7, list = FALSE)
train <- df[index, ]
test <- df[-index, ]

# Model 1: SVM
svm_model <- svm(Kualitas ~ ., data = train, kernel = "radial")
svm_pred <- predict(svm_model, newdata = test)

# Model 2: KNN (k = 5)
knn_pred <- knn(train = train[, -9], test = test[, -9],
                cl = train$Kualitas, k = 5)

# Model 3: Decision Tree
dt_model <- rpart(Kualitas ~ ., data = train, method = "class")
dt_pred <- predict(dt_model, test, type = "class")

# Fungsi evaluasi metrik
get_conf_matrix <- function(pred, actual, model) {
  cm <- confusionMatrix(pred, actual)
  m <- cm$table
  data.frame(
    Model = model,
    TN = m[1,1], FP = m[1,2],
    FN = m[2,1], TP = m[2,2],
    Akurasi = round(cm$overall["Accuracy"] * 100, 2)
  )
}

svm_metrics <- get_conf_matrix(svm_pred, test$Kualitas, "SVM")
knn_metrics <- get_conf_matrix(knn_pred, test$Kualitas, "KNN")
dt_metrics  <- get_conf_matrix(dt_pred, test$Kualitas, "Decision Tree")

all_metrics <- rbind(svm_metrics, knn_metrics, dt_metrics)
print(all_metrics)

##                   Model TN FP FN TP Akurasi
## Accuracy            SVM  1  2 19 37   64.41
## Accuracy1           KNN  0  2 20 37   62.71
## Accuracy2 Decision Tree  7  9 13 30   62.71

# Simpan hasil prediksi
results <- data.frame(
  Actual = test$Kualitas,
  SVM = svm_pred,
  KNN = knn_pred,
  DecisionTree = dt_pred
)

# Pastikan faktor berurutan: OK = negatif, Defect = positif
results$Actual <- factor(results$Actual, levels = c("OK", "Defect"))
results$SVM <- factor(results$SVM, levels = c("OK", "Defect"))
results$KNN <- factor(results$KNN, levels = c("OK", "Defect"))
results$DecisionTree <- factor(results$DecisionTree, levels = c("OK", "Defect"))

# Fungsi visualisasi
plot_model <- function(method, title) {
  ggplot(results, aes(x = Actual, fill = .data[[method]])) +
    geom_bar(position = "dodge") +
    geom_text(stat = "count", aes(label = after_stat(count)),
              position = position_dodge(0.8), vjust = -0.5) +
    labs(title = title, x = "Aktual", y = "Frekuensi") +
    scale_fill_manual(values = c("OK" = "blue", "Defect" = "red"))
}

# Plot
svm_plot <- plot_model("SVM", "Prediksi SVM")
knn_plot <- plot_model("KNN", "Prediksi KNN (k = 5)")
dt_plot  <- plot_model("DecisionTree", "Prediksi Decision Tree")

grid.arrange(svm_plot, knn_plot, dt_plot, ncol = 3)