Python数据科学-技术详解与商业实践-第七讲作业（本次作业希望大家跟帖讨论）

#%%
import os
import pandas as pd

os.chdir(r'D:\Learningfile\天善学院\280_Ben_八大直播八大案例配套课件\提交-第七讲：使用集成算法建立个人银行反欺诈模型\作业')
data=pd.read_excel('FRAUD_TRAIN_Samp.xlsx')
#%%
data.target.value_counts()
#%%
data.F81.value_counts()
#%%
data.F74.value_counts()
#%%
data.F75.value_counts()
#%%
data.F80.value_counts()
#%%
import numpy as np

dict_F81={'无等级':0,
np.nan:0,
'平衡':1,
'稳健':1,
'成长':1,
'进取':1,
'保守':1}

data['F81']=data['F81'].map(dict_F81)

#%%
dict_F75={'M':1,
'F':0,
np.nan:2}
data.F75=data.F75.map(dict_F75)

#%%缺失值处理
data1=data.copy()

data1=data1.drop(['csr_id','F74','F80'],axis=1)

data1=data1.fillna(0)

#%%
from sklearn.cross_validation import train_test_split
data_X=data1.iloc[:,1:]
data_y=data1.iloc[:,0]
train_X,test_X,train_y,test_y=train_test_split(data_X,data_y,test_size=0.4, train_size=0.6, random_state=12345)

#%%

#采样
from imblearn.combine import SMOTETomek
kos = SMOTETomek(random_state=0) # 综合采样
X_kos, y_kos = kos.fit_sample(train_X, train_y)

#%%
#决策树建模
from sklearn.tree import DecisionTreeClassifier
from sklearn import metrics
from sklearn.model_selection import GridSearchCV

clf = DecisionTreeClassifier(criterion='gini', random_state=1234)
param_grid = {'max_depth':[4,12,13,14], 'max_leaf_nodes':[4,12,30,31,32]}
cv = GridSearchCV(clf, param_grid=param_grid, scoring='f1')

cv.fit(X_kos, y_kos)

predict_test = cv.predict(test_X)
test_proba = cv.predict_proba(test_X)[:, 1]
#%%

from sklearn import metrics
fpr_test, tpr_test, th_test = metrics.roc_curve(test_y, test_proba)
print('AUC = %6.4f' %metrics.auc(fpr_test, tpr_test))

#print(metrics.classification_report(test_y, predict_test))
#%%
cv.best_params_

#%%
#神经网络建模
#标准化
from sklearn.preprocessing import MinMaxScaler

scaler = MinMaxScaler()
scaler.fit(train_X)

scaled_train_data = scaler.transform(train_X)
scaled_test_data = scaler.transform(test_X)
# In[5]:
from sklearn.neural_network import MLPClassifier

mlp = MLPClassifier(hidden_layer_sizes=(10,),
activation='logistic', alpha=0.1, max_iter=1000)

mlp.fit(scaled_train_data, train_y)
mlp
#%%
predict_test2 = mlp.predict(scaled_test_data)
test_proba2 = mlp.predict_proba(scaled_test_data)[:, 1]

from sklearn import metrics
fpr_test2, tpr_test2, th_test2 = metrics.roc_curve(test_y, test_proba2)
print('AUC = %6.4f' %metrics.auc(fpr_test2, tpr_test2))
#print(metrics.classification_report(test_y, predict_test2))
#%%
##RF建模
import sklearn.ensemble as ensemble

param_grid = {
'criterion':['entropy','gini'],
'max_depth':[4,5,6,7],
'n_estimators':[24,26,28], #决策树个数-随机森林特有参数
'max_features':[0.02,0.04,0.4], #每棵决策树使用的变量占比-随机森林特有参数
'min_samples_split':[74,75,78]
}

rfc = ensemble.RandomForestClassifier()
rfccv = GridSearchCV(estimator=rfc, param_grid=param_grid, scoring='roc_auc', cv=4,n_jobs=-1)
rfccv.fit(train_X, train_y)
test_est3 = rfccv.predict(test_X)

fpr_test3,tpr_test3,th_test3=metrics.roc_curve(test_y,test_est3)
print('decision tree ROC')
print('ROC=%.4f' %metrics.auc(fpr_test3,tpr_test3))
#%%
rfccv.best_params_

#%%
#Adaboost算法
param_grid = {
#'base_estimator':['DecisionTreeClassifier'],
'learning_rate':[0.1,0.3,0.5,0.7,1]
}

abc = ensemble.AdaBoostClassifier(n_estimators=100,algorithm='SAMME')
abccv = GridSearchCV(estimator=abc, param_grid=param_grid, scoring='roc_auc', cv=4,n_jobs=-1)
abccv.fit(train_X, train_y)
test_est4 = abccv.predict(test_X)
print("abc classifier accuracy:")
print(metrics.classification_report(test_y,test_est4))

fpr_test4,tpr_test4,th_test4=metrics.roc_curve(test_y,test_est4)
print('decision tree ROC')
print('ROC=%.4f' %metrics.auc(fpr_test4,tpr_test4))
#%%
abccv.best_params_

# In[ ]:
#GBDT
param_grid = {
'loss':['deviance','exponential'],
'learning_rate':[0.1,0.3,0.8],
'n_estimators':[26,28,30,32], #决策树个数-GBDT特有参数
'max_depth':[4,5,6], #单棵树最大深度-GBDT特有参数
'min_samples_split':[2,15,20,25]
}

gbc = ensemble.GradientBoostingClassifier()
gbccv = GridSearchCV(estimator=gbc, param_grid=param_grid, scoring='roc_auc', cv=4,n_jobs=-1)
gbccv.fit(train_X, train_y)
test_est5 = gbccv.predict(test_X)
print("gradient boosting accuracy:")
print(metrics.classification_report(test_y,test_est5))

fpr_test5,tpr_test5,th_test5=metrics.roc_curve(test_y,test_est5)
print('decision tree ROC')
print('ROC=%.4f' %metrics.auc(fpr_test5,tpr_test5))
#%%
gbccv.best_params_
#%%#GBDT
param_grid = {
'loss':['deviance','exponential'],
'learning_rate':[0.05,0.1,0.2,0.8],
'n_estimators':[26,30,70,80,90], #决策树个数-GBDT特有参数
'max_depth':[3,4,5], #单棵树最大深度-GBDT特有参数
'min_samples_split':[2,3]
}

gbc = ensemble.GradientBoostingClassifier(random_state=12)
gbccv = GridSearchCV(estimator=gbc, param_grid=param_grid, scoring='roc_auc', cv=4,n_jobs=-1)
gbccv.fit(train_X, train_y)
test_est5 = gbccv.predict(test_X)
print("gradient boosting accuracy:")
print(metrics.classification_report(test_y,test_est5))

fpr_test5,tpr_test5,th_test5=metrics.roc_curve(test_y,test_est5)
print('decision tree ROC')
print('ROC=%.4f' %metrics.auc(fpr_test5,tpr_test5))

gbccv.best_params_
#%%
#PR曲线

import matplotlib.pyplot as plt

plt.figure(figsize=[6,6])
plt.plot(tpr_test2,1-fpr_test2)
plt.ylabel('recall')
plt.xlabel('precisoin')
plt.title('pr curve')
plt.show()

# coding: utf-8

# #### 1、背景介绍：
#
# C银行信用卡中心在对欺诈风险和反欺诈技术作了充分研究之后，融合内外部数据，建立以评分模型为支撑的欺诈识别和防范系统，以满足精细化管理的需要。本次作业根据提供的数据（“FRAUD_TRAIN_Samp.csv”，引用自陈春宝等出版的《SAS金融数据挖掘与建模》）建立信用卡申请反欺诈模型。
#
# #### 2、主要变量说明如下：
#
# #无-组合算法为黑箱模型，无需知道变量含义
#
# #### 3、作业安排：
#
# - 3.1 基础知识：
#
# 1）比较逻辑回归、决策树、神经网络、组合算法的适用场景和优缺点。
#
# - 3.2 案例解答步骤如下：
#
# 1）使用决策树、神经网络、组合算法建立反欺诈模型，比较三个模型的表现。
#
# 2)自学绘制PR曲线：横轴为精确度(Precise)，纵轴为召回率(Recall)。

# In[34]:


import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from scipy import stats
from sklearn.tree import DecisionTreeClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import GridSearchCV
from sklearn import metrics
from sklearn.neural_network import MLPClassifier

get_ipython().run_line_magic('matplotlib', 'inline')
sample=pd.read_excel('samp.xlsx',encoding='utf-8')


# In[35]:
##观测样本目标labeld的分布情况
# In[37]:
sns.distplot(sample.target,kde=True,fit=stats.norm)


# In[38]:
sample_1=sample.loc[sample.iloc[:,1]==1,:]
sample_0=sample.loc[sample.iloc[:,1]==0,:]


# In[39]:

# In[40]:
#初步数据清洗

##数据预处理F81 F74 F75
from pylab import mpl
mpl.rcParams['font.sans-serif']=['SimHei']#指定默认字体
mpl.rcParams['axes.unicode_minus']=False
sample.F81=sample.F81.fillna('缺失')
sample.F81.value_counts().plot(kind='bar')
dict81={'无等级':1,'缺失':2,'平衡':3,'稳健':4,'成长':5,'进取':6,'保守':7}
sample.F81=sample.F81.map(dict81)
sample.F74.value_counts().plot(kind='bar')
dict74={'lvyou':1,'feiji':2,'tielu':3,'xchen':4,'JIUDIAN':5,'quna':6,'JYY':7}
sample.F74=sample.F81.map(dict74)


# In[41]:


sample.F75=sample.F75.fillna('O')
sample.F75.unique()


# In[42]:


sample.F75.value_counts().plot(kind='bar')
dict75={'M':0,'F':1,'O':2}
sample.F75=sample.F75.map(dict75)


# In[43]:


#针对空值转化为0,针对非数字型记录其索引
cloindex=[]
for i in range(245):
sample.iloc[:,i+2]=sample.iloc[:,i+2].fillna(0)
sample=sample.drop('F80',1)


# In[44]:

# In[45]:


sample.target.value_counts().plot(kind='bar')


# #### 使用决策树构建模型

# In[46]:


from sklearn import cross_validation

X_data=sample.iloc[:,2:]
y_data=sample.iloc[:,1]


# In[47]:


X_train,X_test,y_train,y_test=cross_validation.train_test_split(
X_data,y_data,test_size=0.35,random_state=1
)


# In[48]:


regr=DecisionTreeClassifier()
regr.fit(X_train,y_train)
print('Training score:%f'%(regr.score(X_train,y_train)))
print('Testing score:%f'%(regr.score(X_test,y_test)))


# In[49]:


result_train=regr.predict(X_train)
result_train_proba=regr.predict_proba(X_train)[:,1]

result_test=regr.predict(X_test)
result_test_proba=regr.predict_proba(X_test)[:,1]

pd.DataFrame({
'y_test':y_test,
'result_test':result_test,
'result_test_proba':result_test_proba
}).T


# In[61]:


import sklearn.metrics as metrics
print(metrics.confusion_matrix(y_test,result_test,labels=[0,1]))
print(metrics.classification_report(y_test,result_test))
#print(pd.DataFrame(list(zip(sample.columns,regr.feature_importances_))))




# In[56]:


from sklearn.model_selection import GridSearchCV
from sklearn import metrics


# In[58]:


param_grid={
'max_depth':[2,3,4,5,6,7,8],
'min_samples_split':[4,8,12,16,20,24,28]
}
clf=DecisionTreeClassifier(criterion='entropy')
clfcv=GridSearchCV(estimator=clf,param_grid=param_grid,
scoring='roc_auc',cv=4)
clfcv.fit(X_train,y_train)


# In[158]:


mlp=MLPClassifier(hidden_layer_sizes=(10,),
activation='logistic',
alpha=0.1,
max_iter=100000)
mlp.fit(X_train,y_train)
mlp


# In[144]:


train_predict=mlp.predict()


# In[164]:


from sklearn.model_selection import GridSearchCV
from sklearn import metrics
param_grid={
'max_iter':[1000,2000,3000,4000,5000],
'hidden_layer_sizes':[(10,),(15,),(20,),(5,5)],
'activation':['identity','logistic','tanh','relu'],
'alpha':[0.001,0.01,0.1,0.2,0.4,1,10]
}
mlp=MLPClassifier()
gcv=GridSearchCV(estimator=mlp,param_grid=param_grid,
scoring='roc_auc',cv=4,n_jobs=1)
gcv.fit(X_train,y_train)
gcv.best_params_
# In[51]:


import seaborn as sns
red,blue=sns.color_palette('Set1',2)
sns.distplot(result_test_proba[y_test==1],kde=False,bins=15,color=red)
sns.distplot(result_test_proba[y_test==0],kde=False,bins=15,color=blue)
plt.show()


# In[59]:


#fpr_test,tpr_test,th_test=metrics.roc_curve(result_test,result_test_proba)
#fpr_train,tpr_train,th_train=metrics.roc_curve(result_train,result_train_proba)
fpr_test,tpr_test,th_test=metrics.roc_curve(y_test,result_test)
fpr_train,tpr_train,th_train=metrics.roc_curve(y_train,result_train)
plt.figure(figsize=[6,6])
plt.plot(fpr_test,tpr_test,color='blue')
plt.plot(fpr_train,tpr_train,color='red')
plt.title('ROC curve')
plt.show()


# In[62]:


p_test,r_test,_=metrics.precision_recall_curve(y_test,result_test)
p_train,r_train,_=metrics.precision_recall_curve(y_train,result_train)
plt.figure(figsize=[6,6])
plt.plot(r_test,p_test,color='blue')
plt.plot(r_train,p_train,color='red')
plt.title('P-R 曲线')
plt.show()

模型效果如何?不使用超额抽样的方法，是否可以得到理想的模型效果？参数调优的策略是什么？单纯的靠网格搜索法是否可以得到合理的超参数？

只是用了一种调整抽样的方法，这个方法是不是你认为最好的抽样方法？思考一下如何通过更换抽样方式推测样本的空间分布情况。