scikit-learn機器學習(二)邏輯迴歸進行二分類(垃圾郵件分類),二分類性能指標，畫ROC曲線，計算acc,recall,presicion,f1

數據來自UCI機器學習倉庫中的垃圾信息數據集

數據可從http://archive.ics.uci.edu/ml/datasets/sms+spam+collection下載

 

轉成csv載入數據

import matplotlib
matplotlib.rcParams['font.sans-serif']=[u'simHei']
matplotlib.rcParams['axes.unicode_minus']=False
import pandas as pd
import numpy as np
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model.logistic import LogisticRegression
from sklearn.model_selection import train_test_split,cross_val_score

df = pd.read_csv('data/SMSSpamCollection.csv',header=None)
print(df.head)

print("垃圾郵件個數：%s" % df[df[0]=='spam'][0].count())
print("正常郵件個數：%s" % df[df[0]=='ham'][0].count())

垃圾郵件個數：747
正常郵件個數：4825

 

建立TfidfVectorizer實例，將訓練文本和測試文本都進行轉換

X = df[1].values
y = df[0].values
X_train_raw,X_test_raw,y_train,y_test=train_test_split(X,y)
vectorizer = TfidfVectorizer()
X_train = vectorizer.fit_transform(X_train_raw)
X_test = vectorizer.transform(X_test_raw)

 

創建邏輯迴歸模型訓練和預測

LR = LogisticRegression()
LR.fit(X_train,y_train)
predictions = LR.predict(X_test)
for i,prediction in enumerate(predictions[:5]):
    print("預測爲 %s ,信件爲 %s" % (prediction,X_test_raw[i]))

 

預測爲 ham ,信件爲 Send to someone else :-)
預測爲 ham ,信件爲 Easy ah?sen got selected means its good..
預測爲 ham ,信件爲 Sorry da. I gone mad so many pending works what to do.
預測爲 ham ,信件爲 What not under standing.
預測爲 spam ,信件爲 SIX chances to win CASH! From 100 to 20,000 pounds txt> CSH11 and send to 87575. Cost 150p/day, 6days, 16+ TsandCs apply Reply HL 4 info

 

二元分類性能指標：混淆矩陣

# In[2]二元分類分類指標
from sklearn.metrics import confusion_matrix
import matplotlib.pyplot as plt
# predictions 與 y_test
confusion_matrix = confusion_matrix(y_test,predictions)
print(confusion_matrix)
plt.matshow(confusion_matrix)
plt.title("混淆矩陣")
plt.colorbar()
plt.ylabel("真實值")
plt.xlabel("預測值")
plt.show()

[[1217    1]
 [  52  123]]

 

準確率，召回率，精準率，F1值

# In[3] 給出 precision    recall  f1-score   support
from sklearn.metrics import classification_report
print(classification_report(y_test,predictions))

from sklearn.metrics import roc_curve,auc
# 準確率
scores =  cross_val_score(LR,X_train,y_train,cv=5)
print("準確率爲: ",scores)
print("平均準確率爲: ",np.mean(scores))

# 有時必需要將標籤轉爲數值
from sklearn.preprocessing import LabelEncoder
class_le = LabelEncoder()
y_train_n = class_le.fit_transform(y_train)
y_test_n = class_le.fit_transform(y_test)

# 精準率
precision =  cross_val_score(LR,X_train,y_train_n,cv=5,scoring='precision')
print("平均精準率爲: ",np.mean(precision))
# 召回率
recall =  cross_val_score(LR,X_train,y_train_n,cv=5,scoring='recall')
print("平均召回率爲: ",np.mean(recall))   
# F1值
f1 =  cross_val_score(LR,X_train,y_train_n,cv=5,scoring='f1')
print("平均F1值爲: ",np.mean(f1))  

準確率爲:  [0.96654719 0.95459976 0.95449102 0.9508982  0.96047904]
平均準確率爲:  0.9574030433756144
平均精準率爲:  0.9906631114805584
平均召回率爲:  0.6956979405034325
平均F1值爲:  0.8162874707978786

 

畫出ROC曲線，AUC爲ROC曲線如下部分的面積

# In[4] ROC曲線 y_test_n爲數值
predictions_pro = LR.predict_proba(X_test)
false_positive_rate, recall, thresholds = roc_curve(y_test_n,predictions_pro[:,1])
roc_auc = auc(false_positive_rate, recall)
plt.title("受試者操做特徵曲線(ROC)")
plt.plot(false_positive_rate, recall, 'b', label='AUC = % 0.2f' % roc_auc)
plt.legend(loc='lower right')
plt.plot([0,1],[0,1],'r--')
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.0])
plt.xlabel('假陽性率')
plt.ylabel('召回率')
plt.show()
    

 

 全部代碼：

# -*- coding: utf-8 -*-
import matplotlib
matplotlib.rcParams['font.sans-serif']=[u'simHei']
matplotlib.rcParams['axes.unicode_minus']=False
import pandas as pd
import numpy as np
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model.logistic import LogisticRegression
from sklearn.model_selection import train_test_split,cross_val_score

df = pd.read_csv('data/SMSSpamCollection.csv',header=None)
print(df.head)

print("垃圾郵件個數：%s" % df[df[0]=='spam'][0].count())
print("正常郵件個數：%s" % df[df[0]=='ham'][0].count())

# In[1]
X = df[1].values
y = df[0].values
X_train_raw,X_test_raw,y_train,y_test=train_test_split(X,y)
vectorizer = TfidfVectorizer()
X_train = vectorizer.fit_transform(X_train_raw)
X_test = vectorizer.transform(X_test_raw)

LR = LogisticRegression()
LR.fit(X_train,y_train)
predictions = LR.predict(X_test)
for i,prediction in enumerate(predictions[:5]):
    print("預測爲 %s ,信件爲 %s" % (prediction,X_test_raw[i]))
    


# In[2]二元分類分類指標
from sklearn.metrics import confusion_matrix
import matplotlib.pyplot as plt
# predictions 與 y_test
confusion_matrix = confusion_matrix(y_test,predictions)
print(confusion_matrix)
plt.matshow(confusion_matrix)
plt.title("混淆矩陣")
plt.colorbar()
plt.ylabel("真實值")
plt.xlabel("預測值")
plt.show()

# In[3] 給出 precision    recall  f1-score   support
from sklearn.metrics import classification_report
print(classification_report(y_test,predictions))

from sklearn.metrics import roc_curve,auc
# 準確率
scores =  cross_val_score(LR,X_train,y_train,cv=5)
print("準確率爲: ",scores)
print("平均準確率爲: ",np.mean(scores))

# 必需要將標籤轉爲數值
from sklearn.preprocessing import LabelEncoder
class_le = LabelEncoder()
y_train_n = class_le.fit_transform(y_train)
y_test_n = class_le.fit_transform(y_test)

# 精準率
precision =  cross_val_score(LR,X_train,y_train_n,cv=5,scoring='precision')
print("平均精準率爲: ",np.mean(precision))
# 召回率
recall =  cross_val_score(LR,X_train,y_train_n,cv=5,scoring='recall')
print("平均召回率爲: ",np.mean(recall))   
# F1值
f1 =  cross_val_score(LR,X_train,y_train_n,cv=5,scoring='f1')
print("平均F1值爲: ",np.mean(f1))  

# In[4] ROC曲線 y_test_n爲數值
predictions_pro = LR.predict_proba(X_test)
false_positive_rate, recall, thresholds = roc_curve(y_test_n,predictions_pro[:,1])
roc_auc = auc(false_positive_rate, recall)
plt.title("受試者操做特徵曲線(ROC)")
plt.plot(false_positive_rate, recall, 'b', label='AUC = % 0.2f' % roc_auc)
plt.legend(loc='lower right')
plt.plot([0,1],[0,1],'r--')
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.0])
plt.xlabel('假陽性率')
plt.ylabel('召回率')
plt.show()