机器学习06

朴素贝叶斯算法—预测新闻数据

from sklearn.datasets import fetch_20newsgroups
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.naive_bayes import MultinomialNB
from sklearn.model_selection import train_test_split
def pusubeiyesi():
    news=fetch_20newsgroups(subset="all")  #获取数据


    #进行数据分割
    x_train,x_test,y_train,y_test=train_test_split(news.data,news.target,test_size=0.25)
    # print("训练集特征值：",x_train)

    #进行特征抽取
    tf=TfidfVectorizer()

    #以训练集当中的词的列表进行每篇文章的重要性统计
    x_train=tf.fit_transform(x_train)
    # print(tf.get_feature_names_out())
    x_test=tf.transform(x_test)
    #进行朴素贝叶斯算法的预测
    mlt=MultinomialNB(alpha=1.0)


    print(x_train.toarray())
    mlt.fit(x_train,y_train)
    y_predict=mlt.predict(x_test)
    print("预测文章类别为：",y_predict)
    #得出准确率
    print("准确率为：",mlt.score(x_test,y_test))


    return None

pusubeiyesi()

运行结果：

[[0. 0. 0. ... 0. 0. 0.]
 [0. 0. 0. ... 0. 0. 0.]
 [0. 0. 0. ... 0. 0. 0.]
 ...
 [0. 0. 0. ... 0. 0. 0.]
 [0. 0. 0. ... 0. 0. 0.]
 [0. 0. 0. ... 0. 0. 0.]]
预测文章类别为： [12  3 11 ...  9 15 17]
准确率为： 0.8491086587436333

Process finished with exit code 0

朴素贝叶斯算法优缺点

精确率与召回率

API：

from sklearn.datasets import fetch_20newsgroups
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.naive_bayes import MultinomialNB
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report
def pusubeiyesi():
    news=fetch_20newsgroups(subset="all")  #获取数据


    #进行数据分割
    x_train,x_test,y_train,y_test=train_test_split(news.data,news.target,test_size=0.25)
    # print("训练集特征值：",x_train)

    #进行特征抽取
    tf=TfidfVectorizer()

    #以训练集当中的词的列表进行每篇文章的重要性统计
    x_train=tf.fit_transform(x_train)
    # print(tf.get_feature_names_out())
    x_test=tf.transform(x_test)
    #进行朴素贝叶斯算法的预测
    mlt=MultinomialNB(alpha=1.0)


    print(x_train.toarray())
    mlt.fit(x_train,y_train)
    y_predict=mlt.predict(x_test)
    print("预测文章类别为：",y_predict)
    #得出准确率
    print("准确率为：",mlt.score(x_test,y_test))
    print("每个类别的精确率和召回率：", classification_report(y_test, y_predict, target_names=news.target_names))


    return None

pusubeiyesi()

运行结果（添加后）：
每个类别的精确率和召回率：                           precision    recall  f1-score   support

             alt.atheism       0.85      0.72      0.78       209
           comp.graphics       0.88      0.76      0.82       234
 comp.os.ms-windows.misc       0.81      0.85      0.83       233
comp.sys.ibm.pc.hardware       0.76      0.84      0.80       257
   comp.sys.mac.hardware       0.93      0.84      0.88       254
          comp.windows.x       0.94      0.87      0.90       245
            misc.forsale       0.95      0.71      0.82       266
               rec.autos       0.95      0.89      0.92       266
         rec.motorcycles       0.91      0.96      0.93       245
      rec.sport.baseball       0.90      0.94      0.92       232
        rec.sport.hockey       0.94      0.98      0.96       239
               sci.crypt       0.73      0.98      0.84       249
         sci.electronics       0.92      0.80      0.86       244
                 sci.med       0.91      0.93      0.92       209
               sci.space       0.91      0.92      0.92       254
  soc.religion.christian       0.54      0.98      0.70       249
      talk.politics.guns       0.80      0.95      0.87       236
   talk.politics.mideast       0.91      0.98      0.94       231
      talk.politics.misc       0.99      0.62      0.77       200
      talk.religion.misc       1.00      0.17      0.29       160

                accuracy                           0.85      4712
               macro avg       0.88      0.84      0.83      4712
            weighted avg       0.87      0.85      0.84      4712


Process finished with exit code 0

依次是精确率，召回率

模型的选择与调优

交叉验证：

所有数据分成n等分

网格搜索（调参数）：

from sklearn.model_selection import GridSearchCV

 Knn=KNeighborsClassifier()
    gc=GridSearchCV(Knn,param_grid={"n_neighbors":[1,3,5]},cv=2)
    gc.fit(x_train,y_train)
    #预测准确率
    print("在测试集的准确率：",gc.score(x_test))
    #把x_train分为验证集和训练集
    print("在交叉验证中最好的结果：",gc.best_score_)
    print("在交叉验证中最好的参数模型：",gc.best_estimator_)
    print("每次交叉验证后的验证集准确率和训练集准确率",gc.cv_results_)