Scikit-Learn: Machine Learning in Python

【问题描述】

【解决思路】

第一步：先建立一个样本容量为2000、特征数量为20的数据集。

X,y = datasets.make_classification(n_samples=2000, n_features=20)

关于sklearn的数据加载可参看https://blog.youkuaiyun.com/sa14023053/article/details/52086695。

第二步：使用10折交叉检验对数据集进行划分。

kf = cross_validation.KFold(2000, n_folds=10, shuffle=True)  

关于交叉检验可以参看https://blog.youkuaiyun.com/u010414589/article/details/51145341。

第三步：训练算法

这里用的的算法有三种：

（1）朴素贝叶斯分类算法（GaussianNB类）

clf1 = GaussianNB()  
clf1.fit(X_train, y_train)  
pred1 = clf1.predict(X_test)  

关于朴素贝叶斯分类算法的介绍可以参看https://www.cnblogs.com/pinard/p/6074222.html。

（2）支持向量机（SVM）分类算法（rbf核，参数C取1e-1）

clf2 = SVC(C=1e-01, kernel='rbf', gamma=0.1)  
clf2.fit(X_train, y_train)  
pred2 = clf2.predict(X_test) 

（3）随机森林分类算法

clf3 = RandomForestClassifier(n_estimators=6)  
clf3.fit(X_train, y_train)  
pred3 = clf3.predict(X_test)

第四步：评估性能

acc = metrics.accuracy_score(y_test, pred)
f1 = metrics.f1_score(y_test, pred)  
auc = metrics.roc_auc_score(y_test, pred) 

【代码实现】

from sklearn import datasets  
from sklearn import cross_validation  
from sklearn.naive_bayes import GaussianNB  
from sklearn.svm import SVC  
from sklearn.ensemble import RandomForestClassifier  
from sklearn import metrics  
  
# Step1: Create a classification dataset
X,y = datasets.make_classification(n_samples=2000, n_features=20)  
print("X:\n", X)  
print("y:\n", y)  


# Step2: Split the dataset using 10-fold cross validation  
kf = cross_validation.KFold(2000, n_folds=10, shuffle=True)  
for train_index, test_index in kf:  
    X_train, X_test = X[train_index], X[test_index]  
    y_train, y_test = y[train_index], y[test_index]  
  
print("X_train:\n", X_train)  
print("X_test:\n", X_test)  
print("y_train:\n", y_train)  
print("y_test:\n", y_test)  


# Step3: Train the algorithms  
print("\n\nAlgorithm training:")  
# GaussianNB  
print("Gaussian:")  
clf1 = GaussianNB()  
clf1.fit(X_train, y_train)  
pred1 = clf1.predict(X_test)  
print("X_predict:\n", pred1)  
print("y_test:\n", y_test)  
  
# SVC
print("\nSVC:")  
clf2 = SVC(C=1e-01, kernel='rbf', gamma=0.1)  
clf2.fit(X_train, y_train)  
pred2 = clf2.predict(X_test)  
print("X_predict:\n", pred2)  
print("y_test:\n", y_test)  
  
# RandomForestClassifier  
print("\nRandomForestClassifier:")  
clf3 = RandomForestClassifier(n_estimators=6)  
clf3.fit(X_train, y_train)  
pred3 = clf3.predict(X_test)  
print("X_predict:\n", pred3)  
print("y_test:\n", y_test)  


# Step4: Evaluate the cross-validated performance   
print("\n\nPerformance evaluation:")
# GaussianNB
print("Gaussian:")  
acc = metrics.accuracy_score(y_test, pred1)  
print("Accuracy: ",acc)  
f1 = metrics.f1_score(y_test, pred1)  
print("F1-score: ", f1)  
auc = metrics.roc_auc_score(y_test, pred1)  
print("AUC ROC: ", auc)  

# SVC
print("\nSVC:")  
acc = metrics.accuracy_score(y_test, pred2)  
print("Accuracy: ", acc)  
f1 = metrics.f1_score(y_test, pred2)  
print("F1-score: ", f1)  
auc = metrics.roc_auc_score(y_test, pred2)  
print("AUC ROC: ", auc)  

# RandomForestClassifier
print("\nRandomForestClassifier:")  
acc = metrics.accuracy_score(y_test, pred3)  
print("Accuracy: ", acc)  
f1 = metrics.f1_score(y_test, pred3)  
print("F1-score: ", f1)  
auc = metrics.roc_auc_score(y_test, pred3)  
print("AUC ROC: ", auc)  

【运行结果】

多次采用三种算法对随机生成的数据集进行预测，性能评估如下：

    

    

从多次的运行结果可以看出，多数情况下随机森林算法的性能比另外两个要好。