《机器学习实战》决策树的应用

 

 

课本中给出的是一个预测隐形眼镜的例子。

数据集样式如下：

young    myope    no    reduced    no lenses
young    myope    no    normal    soft
young    myope    yes    reduced    no lenses
young    myope    yes    normal    hard
young    hyper    no    reduced    no lenses
young    hyper    no    normal    soft
young    hyper    yes    reduced    no lenses
young    hyper    yes    normal    hard
pre    myope    no    reduced    no lenses

 

用sklearn实现的代码如下：

---

import numpy as np
import pandas as pd

#载入数据
labels=['age','prescript','astigmatic','tearRate','class']
feature=['age','prescript','astigmatic','tearRate']
lenses=pd.read_csv('C:/Users/lenovo/Desktop/lenses.txt',names=labels,sep='\t')
print(lenses)


#将数据转换为数值型
from sklearn import preprocessing

lenses_feature=lenses[feature]
le=preprocessing.LabelEncoder()
for col in lenses_feature.columns:
    lenses_feature[col]=le.fit_transform(lenses_feature[col])

print(lenses_feature)


from sklearn.model_selection import train_test_split

#分离出测试集和验证集
X=lenses_feature
y=lenses['class']
X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.3)

#训练决策树模型
from sklearn import tree

dtc=tree.DecisionTreeClassifier().fit(X_train,y_train)

y_pred=dtc.predict(X_test)
print(dtc.predict([[1,1,0,1]]))


#评估模型
from sklearn import metrics

value=list(set(y))
print(metrics.classification_report(y_test,y_pred))

print(metrics.confusion_matrix(y_test,y_pred,labels=value))

#画决策树
import graphviz

feature_name=['age','prescript','astigmatic','tearRate']

dot_data=tree.export_graphviz(dtc,out_file=None
                                ,feature_names= feature_name
                                ,class_names=["no lenses","soft","hard"]
                                ,filled=True
                                ,rounded=True
                               )
graph = graphviz.Source(dot_data)
graph

#探索决策树
print(dtc.feature_importances_)

print([*zip(feature,dtc.feature_importances_)])