# [Q6] Adaptive Boosting

# [Q6-1] Hyperparameter 후보 값들을 명시하고, Validation dataset을 통해 최적의 hyperparameter 값을 찾아보시오.

AdaBoost.trn <- CART.trn
AdaBoost.val <- CART.val
AdaBoost.tst <- CART.tst

mfinallist <- c(50,100,150) # mfinal(반복 횟수)을 이 3가지로 바꿔가며 테스트
val_perf <- matrix(0, length(mfinallist), 7)
colnames(val_perf) <- c("mfinal", "TPR", "Precision", "TNR", "Accuracy", "BCR", "F1-Measure")
for (i in 1:length(mfinallist)) {

  AdaBoost.model <- boosting(earthquateYN~., AdaBoost.trn, boos=TRUE,
                             mfinal=mfinallist[i])

  AdaBoost.prey <- predict(AdaBoost.model, AdaBoost.val[,-69])
  AdaBoost.cfm <- table(AdaBoost.val$earthquateYN, AdaBoost.prey$class)
  # Confusion matrix
  val_perf[i,1] <- mfinallist [i]
  # Record the validation performance
  val_perf[i,2:7] <- perf_eval(AdaBoost.cfm)
}
ada_val_perf <- val_perf