集成学习和代码实现

 1 # 导入算法包以及数据集
 2 from sklearn import neighbors
 3 from sklearn import datasets
 4 from sklearn.ensemble import BaggingClassifier
 5 from sklearn import tree
 6 from sklearn.model_selection import train_test_split
 7 import numpy as np
 8 import matplotlib.pyplot as plt
 9 iris = datasets.load_iris()
10 x_data = iris.data[:,:2]
11 y_data = iris.target
12 def plot(model):
13     # 获取数据值所在的范围
14     x_min, x_max = x_data[:, 0].min() - 1, x_data[:, 0].max() + 1
15     y_min, y_max = x_data[:, 1].min() - 1, x_data[:, 1].max() + 1
16 
17     # 生成网格矩阵
18     xx, yy = np.meshgrid(np.arange(x_min, x_max, 0.02),
19                          np.arange(y_min, y_max, 0.02))
20 
21     z = model.predict(np.c_[xx.ravel(), yy.ravel()])# ravel与flatten类似，多维数据转一维。flatten不会改变原始数据，ravel会改变原始数据
22     z = z.reshape(xx.shape)
23     # 等高线图
24     cs = plt.contourf(xx, yy, z)
25 x_train,x_test,y_train,y_test = train_test_split(x_data, y_data)
26 knn = neighbors.KNeighborsClassifier()
27 knn.fit(x_train, y_train)
28 # 画图
29 plot(knn)
30 # 样本散点图
31 plt.scatter(x_data[:, 0], x_data[:, 1], c=y_data)
32 plt.show()
33 # 准确率
34 print("knn:",knn.score(x_test, y_test))
35 dtree = tree.DecisionTreeClassifier()
36 dtree.fit(x_train, y_train)
37 # 画图
38 plot(dtree)
39 # 样本散点图
40 plt.scatter(x_data[:, 0], x_data[:, 1], c=y_data)
41 plt.show()
42 # 准确率
43 print("dtree:",dtree.score(x_test, y_test))
44 bagging_knn = BaggingClassifier(knn, n_estimators=100)#使用bagging算法训练100组knn分类器
45 # 输入数据建立模型
46 bagging_knn.fit(x_train, y_train)
47 plot(bagging_knn)
48 # 样本散点图
49 plt.scatter(x_data[:, 0], x_data[:, 1], c=y_data)
50 plt.show()
51 print("bagging:",bagging_knn.score(x_test, y_test))
52 bagging_tree = BaggingClassifier(dtree, n_estimators=100)#使用bagging算法训练100组决策树分类器
53 # 输入数据建立模型
54 bagging_tree.fit(x_train, y_train)
55 plot(bagging_tree)
56 # 样本散点图
57 plt.scatter(x_data[:, 0], x_data[:, 1], c=y_data)
58 plt.show()
59 print("bagging_tree",bagging_tree.score(x_test, y_test))

 1 from sklearn import tree
 2 from sklearn.model_selection import train_test_split
 3 from sklearn.ensemble import RandomForestClassifier
 4 import numpy as np
 5 import matplotlib.pyplot as plt
 6 # 载入数据
 7 data = np.genfromtxt("LR-testSet2.txt", delimiter=",")
 8 x_data = data[:,:-1]
 9 y_data = data[:,-1]
10 
11 plt.scatter(x_data[:,0],x_data[:,1],c=y_data)
12 plt.show()
13 x_train,x_test,y_train,y_test = train_test_split(x_data, y_data, test_size = 0.5)
14 def plot(model):
15     # 获取数据值所在的范围
16     x_min, x_max = x_data[:, 0].min() - 1, x_data[:, 0].max() + 1
17     y_min, y_max = x_data[:, 1].min() - 1, x_data[:, 1].max() + 1
18 
19     # 生成网格矩阵
20     xx, yy = np.meshgrid(np.arange(x_min, x_max, 0.02),
21                          np.arange(y_min, y_max, 0.02))
22 
23     z = model.predict(np.c_[xx.ravel(), yy.ravel()])# ravel与flatten类似，多维数据转一维。flatten不会改变原始数据，ravel会改变原始数据
24     z = z.reshape(xx.shape)
25     # 等高线图
26     cs = plt.contourf(xx, yy, z)
27     # 样本散点图
28     plt.scatter(x_test[:, 0], x_test[:, 1], c=y_test)
29     plt.show()
30 dtree = tree.DecisionTreeClassifier()
31 dtree.fit(x_train, y_train)
32 plot(dtree)
33 print("dtree",dtree.score(x_test, y_test))
34 RF = RandomForestClassifier(n_estimators=50)
35 RF.fit(x_train, y_train)
36 plot(RF)
37 print("RandomForest",RF.score(x_test, y_test))

 1 import numpy as np
 2 import matplotlib.pyplot as plt
 3 from sklearn import tree
 4 from sklearn.ensemble import AdaBoostClassifier
 5 from sklearn.tree import DecisionTreeClassifier
 6 from sklearn.datasets import make_gaussian_quantiles
 7 from sklearn.metrics import classification_report
 8 # 生成2维正态分布，生成的数据按分位数分为两类，500个样本,2个样本特征
 9 x1, y1 = make_gaussian_quantiles(n_samples=500, n_features=2,n_classes=2)#默认mean=(0, 0)
10 # 生成2维正态分布，生成的数据按分位数分为两类，400个样本,2个样本特征均值都为3
11 x2, y2 = make_gaussian_quantiles(mean=(3, 3), n_samples=500, n_features=2, n_classes=2)
12 # 将两组数据合成一组数据
13 x_data = np.concatenate((x1, x2))
14 y_data = np.concatenate((y1, - y2 + 1))
15 plt.scatter(x_data[:, 0], x_data[:, 1], c=y_data)
16 plt.show()
17 # 决策树模型
18 model = tree.DecisionTreeClassifier(max_depth=3)
19 
20 # 输入数据建立模型
21 model.fit(x_data, y_data)
22 
23 # 获取数据值所在的范围
24 x_min, x_max = x_data[:, 0].min() - 1, x_data[:, 0].max() + 1
25 y_min, y_max = x_data[:, 1].min() - 1, x_data[:, 1].max() + 1
26 
27 # 生成网格矩阵
28 xx, yy = np.meshgrid(np.arange(x_min, x_max, 0.02),
29                      np.arange(y_min, y_max, 0.02))
30 
31 z = model.predict(np.c_[xx.ravel(), yy.ravel()])# ravel与flatten类似，多维数据转一维。flatten不会改变原始数据，ravel会改变原始数据
32 z = z.reshape(xx.shape)
33 # 等高线图
34 cs = plt.contourf(xx, yy, z)
35 # 样本散点图
36 plt.scatter(x_data[:, 0], x_data[:, 1], c=y_data)
37 plt.show()
38 # 模型准确率
39 print("决策树：",model.score(x_data,y_data))
40 # AdaBoost模型
41 model = AdaBoostClassifier(DecisionTreeClassifier(max_depth=3),n_estimators=10)#决策树深度为3，一共训练十个模型
42 # 训练模型
43 model.fit(x_data, y_data)
44 
45 # 获取数据值所在的范围
46 x_min, x_max = x_data[:, 0].min() - 1, x_data[:, 0].max() + 1
47 y_min, y_max = x_data[:, 1].min() - 1, x_data[:, 1].max() + 1
48 
49 # 生成网格矩阵
50 xx, yy = np.meshgrid(np.arange(x_min, x_max, 0.02),
51                      np.arange(y_min, y_max, 0.02))
52 
53 # 获取预测值
54 z = model.predict(np.c_[xx.ravel(), yy.ravel()])
55 z = z.reshape(xx.shape)
56 # 等高线图
57 cs = plt.contourf(xx, yy, z)
58 # 样本散点图
59 plt.scatter(x_data[:, 0], x_data[:, 1], c=y_data)
60 plt.show()
61 # 模型准确率
62 print("adaboost：",model.score(x_data,y_data))#得分很高

 1 from sklearn import datasets
 2 from sklearn import model_selection
 3 from sklearn.linear_model import LogisticRegression
 4 from sklearn.neighbors import KNeighborsClassifier
 5 from sklearn.tree import DecisionTreeClassifier
 6 from mlxtend.classifier import StackingClassifier # pip install mlxtend
 7 import numpy as np
 8 
 9 # 载入数据集
10 iris = datasets.load_iris()
11 # 只要第1,2列的特征
12 x_data, y_data = iris.data[:, 1:3], iris.target
13 
14 # 定义三个不同的分类器
15 clf1 = KNeighborsClassifier(n_neighbors=1)
16 clf2 = DecisionTreeClassifier()
17 clf3 = LogisticRegression()
18 
19 # 定义一个次级分类器
20 lr = LogisticRegression()
21 sclf = StackingClassifier(classifiers=[clf1, clf2, clf3],
22                           meta_classifier=lr)
23 
24 for clf, label in zip([clf1, clf2, clf3, sclf],
25                       [‘KNN‘, ‘Decision Tree‘, ‘LogisticRegression‘, ‘StackingClassifier‘]):
26     scores = model_selection.cross_val_score(clf, x_data, y_data, cv=3, scoring=‘accuracy‘)
27     print("Accuracy: %0.2f [%s]" % (scores.mean(), label))

 1 from sklearn import datasets
 2 from sklearn import model_selection
 3 from sklearn.linear_model import LogisticRegression
 4 from sklearn.neighbors import KNeighborsClassifier
 5 from sklearn.tree import DecisionTreeClassifier
 6 from sklearn.ensemble import VotingClassifier
 7 import numpy as np
 8 
 9 # 载入数据集
10 iris = datasets.load_iris()
11 # 只要第1,2列的特征
12 x_data, y_data = iris.data[:, 1:3], iris.target
13 
14 # 定义三个不同的分类器
15 clf1 = KNeighborsClassifier(n_neighbors=1)
16 clf2 = DecisionTreeClassifier()
17 clf3 = LogisticRegression()
18 
19 sclf = VotingClassifier([(‘knn‘, clf1), (‘dtree‘, clf2), (‘lr‘, clf3)])
20 
21 for clf, label in zip([clf1, clf2, clf3, sclf],
22                       [‘KNN‘, ‘Decision Tree‘, ‘LogisticRegression‘, ‘VotingClassifier‘]):
23     scores = model_selection.cross_val_score(clf, x_data, y_data, cv=3, scoring=‘accuracy‘)
24     print("Accuracy: %0.2f [%s]" % (scores.mean(), label))