朴素貝葉斯應用：垃圾郵件分類

 

1. 數據准備：收集數據與讀取

2. 數據預處理：處理數據

3. 訓練集與測試集：將先驗數據按一定比例進行拆分。

4. 提取數據特征，將文本解析為詞向量 。

5. 訓練模型：建立模型，用訓練數據訓練模型。即根據訓練樣本集，計算詞項出現的概率P(xi|y)，后得到各類下詞匯出現概率的向量 。 

6. 測試模型：用測試數據集評估模型預測的正確率。

混淆矩陣 ，准確率、精確率、召回率、F值 

7. 預測一封新郵件的類別。

 

#導入nltk數據包
import nltk
from nltk.corpus import stopwords
from nltk.stem import WordNetLemmatizer

#導入包
import csv  
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.metrics import confusion_matrix
from sklearn.metrics import classification_report

text = '''As per your request 'Melle Melle (Oru Minnaminunginte Nurungu Vettam)' has been set as your callertune for all Callers. Press *9 to copy your friends Callertune'''

#進行郵件預處理
def preprocessing(text):
    text=text.decode("utf-8")
    # 分詞
    tokens = [word for sent in nltk.sent_tokenize(text) for word in nltk.word_tokenize(sent)]
    stops = stopwords.words('english'）  #停用詞
    tokens = [token for token in tokens if token not in stops]  #去掉停用詞
    tokens = [token.lower() for token in tokens if len(token) >= 3]     #去掉短於3的詞
   
     #詞性還原                       
    lmtzr = WordNetLemmatizer()   
    tokens = [lmtzr.lemmatize(token) for token in tokens]
     
    #將剩下的詞重新連接成字符串
    preprocessed_text = ' '.join(tokens)
    return preprocessed_text

 
#讀數據
file_path = r'C:\Users\Administrator\Desktop\ems.txt'
ems = open(file_path,'r',encoding='utf-8')
ems_data=[]  
ems_label=[]   #保存
csv_reader=csv.reader(ems,delimiter='\t')

 #將數據分別存入數據列表和目標分類列表                           
for line in csv_reader:
    ems_label.append(line[0])
    ems_data.append(preprocessing(line[1]))
ems.close()

 

#將數據分為訓練集和測試集，再將其向量化
from sklearn.model_selection import train_test_split
x_train,x_test,y_train,y_test=train_test_split(ems_data,ems_target,test_size=0.3,random_state=0,startify=ems_target)
print(len(x_train,len(x_test)))
                            

# 將其向量化
from sklearn.feature_extraction.text import TfidfVectorizer   #建立數據的特征向量
vectorizer=TfidfVectorizer(min_df=2,ngram_range=(1,2),stop_words='english',strip_accents='unicode',norm='l2')
X_train = vectorizer.fit_transform(x_train)
X_test = vectorizer.transform(x_test)

import numpy as np               #觀察向量
a = X_train.toarray()
                            
for i in range(1000):            #輸出不為0的列
    for j in range(5984):
        if a[i,j]!=0:
            print(i,j,a[i,j])
                            

#朴素貝葉斯分類器
from sklearn.navie_bayes import MultinomialNB
clf = MultinomialNB().fit(X_train,y_train)
y_nb_pred = clf.predict(X_test)


# 分類結果顯示
print(y_nb_pred.shape,y_nb_pred) # x-test預測結果
print('nb_confusion_matrix:')
cm = confusion_matrix(y_test,y_nb_pred) #混淆矩陣
print(cm)
print('nb_classification_repert:')
cr = classification_report(y_test,y_nb_pred) # 主要分類指標的文本報告
print(cr)

 
feature_names=vectorizer.get_feature_names() # 出現過的單詞列表
coefs=clf.coef_ # 先驗概率 p(x_ily),6034 feature_log_preb
intercept = clf.intercept_ # P(y),class_log_prior : array,shape(n...
coefs_with_fns=sorted(zip(coefs[0],feature_names)) #對數概率P(x_i|y)與單詞x_i映射


n=10
top=zip(coefs_with_fns[:n],coefs_with_fns[:-(n+1):-1]) #最大的10個與最小的10個單詞
for (coef_1,fn_1),(coef_2,fn_2) in top:
    print('\t%.4f\t%-15s\t\t%.4f\t%-15s' % (coef_1,fn_1,coef_2,fn_2))

                            
#預測一封新郵件的類別。
new_email=['新郵件']
vectorizer(new_email)
clf.predict(new_email)

 

結果：