Syntax Jurnal Classification of Malaria Complication Using CART (Classification and Regression Tree) and Naïve Bayes

INPUT DATA

library(rpart.plot)
library(caret)
library(e1071)
library(corrplot)
library(ggplot2)
library(reshape2)

data <- read.csv("clipboard", header = TRUE)

names(data)[names(data) == "severe_maleria"] <- "severe_malaria"
names(data)[names(data) == "headace"] <- "headache"
names(data)[names(data) == "prostraction"] <- "prostration"

str(data)
summary(data)

PREPROCESSING DATA

```# Ubah target ke faktor data\(severe_malaria <- as.factor(data\)severe_malaria)

Normalisasi age (min-max)

data\(age <- (data\)age - min(data\(age)) / (max(data\)age) - min(data$age))

## THE MEASURE OF CORRELATION

```data_num <- data[, sapply(data, is.numeric)]
corr_matrix <- cor(data_num, use = "complete.obs")

# Tampilkan matriks
corr_matrix

ATTRIBUTE’S CORRELATION

         method = "color",
         type = "full",
         col = colorRampPalette(c("blue","white","red"))(200),
         addCoef.col = "black",
         number.cex = 0.5,
         tl.col = "black",
         tl.cex = 0.7,
         tl.srt = 45)

# Plot pohon CART
rpart.plot(model_cart,
           type = 2,
           extra = 104,
           fallen.leaves = TRUE,
           main = "CART Tree for Severe Malaria Classification")

if(exists("pred_nb")){
  
  # Tabel prediksi
  table_pred <- table(pred_nb)
  
  # Ubah ke persen
  prob_pred <- prop.table(table_pred) * 100
  
  # Gabungkan
  data.frame(Complication = names(prob_pred),
             Percentage = round(prob_pred, 2))
}
if(exists("cm_nb")){
  accuracy <- sum(diag(cm_nb)) / sum(cm_nb)
  accuracy
}
set.seed(123)

results <- data.frame(Attempt = 1:3,
                      Accuracy_CART = NA,
                      Accuracy_NB = NA)

for(i in 1:3){
  
  # Split ulang
  train_index <- createDataPartition(data$severe_malaria, p = 0.7, list = FALSE)
  train_data <- data[train_index, ]
  test_data  <- data[-train_index, ]
  
  # CART
  model_cart <- rpart(severe_malaria ~ ., 
                      data = train_data,
                      method = "class",
                      control = rpart.control(maxdepth = 4))
  
  pred_cart <- predict(model_cart, test_data, type = "class")
  
  acc_cart <- mean(pred_cart == test_data$severe_malaria)

SPLIT DATA CART

train_index <- createDataPartition(data$severe_malaria, 
                                   p = 0.7, 
                                   list = FALSE)

train_data <- data[train_index, ]
test_data  <- data[-train_index, ]

MODEL CART

model_cart <- rpart(severe_malaria ~ ., data = train_data, method = "class", control = rpart.control(maxdepth = 4))

PREDIKSI CART

pred_cart <- predict(model_cart, test_data, type = "class")

DATA UNTUK NAÏVE BAYES

data_nb <- test_data[pred_cart == 1, ]

CONFUSION MATRIX CART

```cm <- table(Predicted = pred_cart, Actual = test_data$severe_malaria)

cm

Hitung metrik

cm <- table(Predicted = pred_cart, Actual = test_data$severe_malaria)

cm TP <- cm[2,2] TN <- cm[1,1] FP <- cm[2,1] FN <- cm[1,2]

precision <- TP / (TP + FP) recall <- TP / (TP + FN) f1 <- 2 * precision * recall / (precision + recall)

Tampilkan hasil

data.frame(Precision = precision, Recall = recall, F1_Score = f1)

##  NAÏVE BAYES (jika data cukup)

```set.seed(123)

results <- data.frame(Attempt = 1:3,
                      Accuracy_CART = NA,
                      Accuracy_NB = NA)

for(i in 1:3){
  
  # Split ulang
  train_index <- createDataPartition(data$severe_malaria, p = 0.7, list = FALSE)
  train_data <- data[train_index, ]
  test_data  <- data[-train_index, ]
  
  # CART
  model_cart <- rpart(severe_malaria ~ ., 
                      data = train_data,
                      method = "class",
                      control = rpart.control(maxdepth = 4))
  
  pred_cart <- predict(model_cart, test_data, type = "class")
  
  acc_cart <- mean(pred_cart == test_data$severe_malaria)
  
  # NB (jika cukup data)
  data_nb <- test_data[pred_cart == 1, ]
  
  if(nrow(data_nb) > 2){
    
    data_nb$complication <- as.factor(data_nb$complication)
    
    train_index_nb <- createDataPartition(data_nb$complication, p = 0.85, list = FALSE)
    
    train_nb <- data_nb[train_index_nb, ]
    test_nb  <- data_nb[-train_index_nb, ]
    
    model_nb <- naiveBayes(complication ~ ., data = train_nb)
    pred_nb <- predict(model_nb, test_nb)
    
    acc_nb <- mean(pred_nb == test_nb$complication)
    
  } else {
    acc_nb <- NA
  }
  
  results$Accuracy_CART[i] <- acc_cart
  results$Accuracy_NB[i]   <- acc_nb
}

results