我从“scikit learn/decision tree/export graphviz”的决策树中得到了错误的类名。程序如下所示：

import matplotlib.pyplot as plt
import matplotlib.image as img
import pydot
from sklearn import tree

digital_table = [[0, 0], [0, 1], [1, 0], [1, 1]]
digital_label = ['zero', 'one', 'two', 'three']
digital_name = ['idx-1', 'idx-2']

digital_tree = tree.DecisionTreeClassifier()
digital_tree.fit(digital_table, digital_label)

with open("digital.dot", 'w') as f:
    f = tree.export_graphviz(digital_tree, 
                            feature_names=digital_name,
                            class_names=digital_label,
                            filled=True, rounded=True,
                            out_file=f)
(graph,) = pydot.graph_from_dot_file("digital.dot")
graph.write_png("digital.png")

plt.imshow(img.imread('digital.png'))
plt.show()

产出如下：

问题在于叶中显示的类名。例如，如果idx-1为1，idx-2为1，则绿色框应标记为“三”。但是，图像显示标签为“一”。有人能发表你的意见吗？