The analysis method of Extra Tree is very similar to Decision Tree. The difference is only when the threshold value of each segmentation is selected by the two algorithms, Decision Tree will select the best segmentation point from all the features, which means that the feature with the best prediction effect after segmentation (the lowest mean square error is selected during regression) is selected for cutting. On the other hand, Extra Tree first randomly selects a segmentation point from each feature, and then selects the feature that can make the best prediction effect (choose the lowest mean square error) for segmentation.

Extra Tree的分析方法和Decision Tree非常相似，其差異僅在兩種演算法進行每一次分割的閥值選定時，Decision Tree會從所有的特徵中選擇最佳分割點，意即選擇分割後預測效果最佳(回歸時選擇均方誤差最低)的特徵進行切割，而Extra Tree則是先從每個特徵中隨機選擇一個分割點，再從其挑選出能使預測效果最佳(選擇均方誤差最低)之特徵進行分割。