pydotplus安装和基本入门

import pydotplus as pdp

%matplotlib inline

from IPython.display import display, Image
graph = pdp.graph_from_dot_data('digraph demo1{ a -> b -> c; c ->a; }')
display(Image(graph.create_png()))

dot = '''
//定义节点属性
  digraph g {
      //==========定义节点关系============
      a->b;
      b->c;
      c->a;
      c->d->e->f;
      d->g;
      e->h;
      //==========定义节点属性============
      //定义a节点为长方形, 样式为填充, 填充颜色为#ABACBA
      a[shape=box,label="Server1\nWebServer",fillcolor="#ABACBA",style=filled];
      //定义b为5边形, 标签为"bb", 样式为填充, 填充色为red
      b[shape=polygon,sides=5,label="bb",style=filled,fillcolor=red];
      //c, 默认为椭圆
      d[shape=circle]; //园
      e[shape=triangle]; //三角形
      f[shape=polygon, sides=4, skew=0.5]; //平行四边形
      g[shape=polygon, distortion=0.5]; //梯形, 上边长
      h[shape=polygon, distortion=-.5]; //梯形, 下边长
  }
'''

graph = pdp.graph_from_dot_data(dot)
# 下面这行用于在notebook中显示图片
Image(graph.create_png())

dot = '''
digraph graphname {
    subgraph clusterA {
        node [shape=plaintext,style=filled];
        1 -> 2 [arrowhead=dot];
        2 -> 3 -> X2 -> 5[arrowhead=tee];
        6;
        7;
        label = "A";
        color=blue
    }
}
'''
graph = pdp.graph_from_dot_data(dot)
# 下面这行用于在notebook中显示图片
Image(graph.create_png())

#用于数据处理和分析的工具包
import pandas as pd
#引入用于数据预处理/特征工程的工具包
from sklearn import preprocessing
#import决策树建模包
from sklearn import tree
from IPython.display import display, Image
adult_data = pd.read_csv('DecisionTree.csv')
feature_columns = [u'workclass', u'education', u'marital-status', u'occupation', u'relationship', u'race', u'gender', u'native-country']
label_column = ['income']
features = adult_data[feature_columns]
label = adult_data[label_column]
features = pd.get_dummies(features)
#初始化一个决策树分类器
clf = tree.DecisionTreeClassifier(criterion='entropy', max_depth=3)
#用决策树分类器拟合数据
clf = clf.fit(features.values, label.values)
dot_data = tree.export_graphviz(clf, 
                                out_file=None, 
                                feature_names=features.columns,
                                class_names = ['<=50k', '>50k'],
                                filled = True,
                                rounded =True
                               )
graph = pdp.graph_from_dot_data(dot_data)
display(Image(graph.create_png()))