Post pruning pohon keputusan spasial untuk klasifikasi kemunculan titik panas

Abstract

The data with noises and outliers to build a decision tree tend to make the size of the decision tree large with low accuracy. This situation is called overfitting. Pruning method is one way to reduce size of the decision tree for increasing its accuracy. This study aims to prune decision trees from previous studies using the post pruning method. The post pruning method gives simple trees with higher accuracies compared to the original trees. The lowest average accuracy of decision tree after post pruning is 0.53% and 28.17% is the highest average accuracy of decision tree after post pruning.