数据分析03---决策树

标签(空格分隔): 数据分析


CART 创建决策树做分类

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
30
31
32
33
# encoding=utf-8
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
from sklearn.tree import DecisionTreeClassifier
from sklearn.datasets import load_iris
import graphviz
from sklearn import tree
import os
os.environ["PATH"] += os.pathsep + 'C:\\Users\\qincf\\AppData\\Local\\Continuum\\anaconda3\Library\\bin\\graphviz'

# 准备数据集
iris=load_iris()
# 获取特征集和分类标识
features = iris.data
labels = iris.target
# 随机抽取 33% 的数据作为测试集,其余为训练集
train_features, test_features, train_labels, test_labels = train_test_split(features, labels, test_size=0.33, random_state=0)
# 创建 CART 分类树
clf = DecisionTreeClassifier(criterion='gini')
# 拟合构造 CART 分类树
clf = clf.fit(train_features, train_labels)
# 用 CART 分类树做预测
test_predict = clf.predict(test_features)
print(test_predict)
# 预测结果与测试集结果作比对
score = accuracy_score(test_labels, test_predict)
print("CART 分类树准确率 %.4lf" % score)

##打印CART
dot_data = tree.export_graphviz(clf,out_file=None)
graph = graphviz.Source(dot_data)
graph
print(graph.view())

分类树如下:

image.png

CART 回归树做预测

1
2
3
4
5
6
7
8
9
10
# encoding=utf-8
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error,mean_absolute_error

from sklearn.tree import DecisionTreeRegressor
from sklearn.datasets import load_boston
import graphviz
from sklearn import tree
import os
os.environ["PATH"] += os.pathsep + 'C:\\Users\\qincf\\AppData\\Local\\Continuum\\anaconda3\Library\\bin\\graphviz'

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
# 准备数据集
boston=load_boston()
# 探索数据
print(boston.feature_names)
# 获取特征集和房价
features = boston.data
prices = boston.target
# 随机抽取 33% 的数据作为测试集,其余为训练集
train_features, test_features, train_price, test_price = train_test_split(features, prices, test_size=0.33)
# 创建 CART 回归树
dtr=DecisionTreeRegressor()
# 拟合构造 CART 回归树
dtr.fit(train_features, train_price)
# 预测测试集中的房价
predict_price = dtr.predict(test_features)
print(test_features)
# 测试集的结果评价
print('回归树二乘偏差均值:', mean_squared_error(test_price, predict_price))
print('回归树绝对值偏差均值:', mean_absolute_error(test_price, predict_price))

本博客所有文章除特别声明外,均采用 CC BY-SA 4.0 协议 ,转载请注明出处!