python绘制决策树_pydotplus 画决策树

from sklearn.datasets import load_iris
from sklearn.tree import DecisionTreeClassifier 
from sklearn.model_selection import train_test_split
from sklearn.tree import export_graphviz
from IPython.display import Image  
import matplotlib.pyplot as plt
import pydotplus

# 1.定义X和y
X = df_churn.iloc[:, :-1]
y = df_churn.Status


# 2. 建立决策树模型
dt_model = DecisionTreeClassifier(max_depth=30, min_samples_split=50, min_samples_leaf=25, 
                                  max_leaf_nodes=100, class_weight='balanced', ccp_alpha=0.0001)
# Fit model to training data
dt_model.fit(X_train, y_train)

# 3. 决策树的可视化
tmp_dot_file = 'decision_tree_tmp.dot'
export_graphviz(dt_model, out_file=tmp_dot_file,filled=True,feature_names=X.columns, class_names=list(set(y)),impurity=False)
with open(tmp_dot_file) as f:
    dot_graph = f.read()
graph = pydotplus.graph_from_dot_data(dot_graph)
graph.write_pdf('example.pdf')    #保存图像为pdf格式
Image(graph.create_png())   #绘制图像为png格式

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

就可以得到这样的结果啦: