八大直播作业-第六讲-Aringrhus

# coding: utf-8



# In[1]:





import numpy as np

import pandas as pd

import os

import seaborn as sns

import matplotlib.pyplot as plt

os.chdir(r"D:\learningFiels\天善常课\061第6讲\提交-第六讲：使用决策树和神经网络预测客户流失倾向\作业")

get_ipython().magic('matplotlib inline')





# In[2]:





marrige = pd.read_csv('date_data2.csv')

marrige.head()





# income-月均收入（元）

# 

# attractive-由婚恋网站评定出的个人魅力值,分值从0-100。

# 

# assets-资产(万元)

# 

# edueduclass-教育等级:1=小学,2=初中;3=高中,4=本科,5=硕士及以上

# 

# Dated-是否相亲成功:1代表成功



# # 数据探索



# In[23]:





print(marrige.info())

marrige.describe()





# In[16]:





sns.boxplot(x='Dated',y='income',data=marrige)





# In[17]:





sns.boxplot(x='Dated',y='attractive',data=marrige)





# In[19]:





sns.boxplot(x='Dated',y='assets',data=marrige)





# In[22]:





marrige.groupby('edueduclass')['Dated'].mean().plot(kind='bar')





# # 初版决策树



# In[25]:





import sklearn.model_selection as cross_validation



target = marrige['Dated']

data = marrige.iloc[:,:-1]



data.head()





# In[26]:





X_train,X_test,y_train,y_test = cross_validation.train_test_split(data,target,train_size=0.7,random_state=123)





# In[31]:





import sklearn.tree as tree





# In[65]:





clf = tree.DecisionTreeClassifier(criterion='entropy',max_depth=8,min_samples_split=3,min_samples_leaf=2)

clf.fit(X_train,y_train)





# In[66]:





train_pre = clf.predict(X_train)

train_pre_p = clf.predict_proba(X_train)[:,1]





# In[67]:





test_pre = clf.predict(X_test)

test_pre_p = clf.predict_proba(X_test)[:,1]





# In[68]:





pd.DataFrame({'test_target':y_test,'test_est':test_pre,'test_est_p':test_pre_p})





# In[69]:





import sklearn.metrics as metrics





# In[70]:





print(metrics.confusion_matrix(y_test,test_pre,labels=[0,1]))





# In[71]:





y_test.sum()





# In[72]:





test_pre.sum()





# In[73]:





print(metrics.classification_report(y_test,test_pre))





# In[74]:





print(pd.DataFrame(list(zip(data.columns,clf.feature_importances_))))





# In[75]:





fpr_test,tpr_test,th_test = metrics.roc_curve(y_test,test_pre)





# In[84]:





print(fpr_test,tpr_test,th_test)





# In[77]:





fpr_train,tpr_train,th_train = metrics.roc_curve(y_train,train_pre)





# In[86]:





print(fpr_train,tpr_train,th_train)





# In[78]:





red, blue = sns.color_palette("Set1", 2)





# In[79]:





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[83]:





print('AUC = %.4f' %metrics.auc(fpr_test, tpr_test))





# # 初版神经网络



# In[4]:





import sklearn.model_selection as cross_validation



target = marrige['Dated']

data = marrige.iloc[:,:-1]



data.head()





# In[5]:





X_train,X_test,y_train,y_test = cross_validation.train_test_split(data,target,train_size=0.7,random_state=123)





# In[8]:





from sklearn.preprocessing import MinMaxScaler



scaler = MinMaxScaler()

scaler.fit(X_train)



scaled_train_data = scaler.transform(X_train)

scaled_test_data = scaler.transform(X_test)





# In[10]:





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, y_train)

mlp





# In[11]:





train_predict = mlp.predict(scaled_train_data)

test_predict = mlp.predict(scaled_test_data)





# In[12]:





# 计算概率

train_proba = mlp.predict_proba(scaled_train_data)[:, 1]  

test_proba = mlp.predict_proba(scaled_test_data)[:, 1]





# In[13]:





from sklearn import metrics



print(metrics.confusion_matrix(y_test, test_predict, labels=[0, 1]))

print(metrics.classification_report(y_test, test_predict))





# In[15]:





mlp.score(scaled_test_data, y_test)





# In[16]:





fpr_test, tpr_test, th_test = metrics.roc_curve(y_test, test_proba)

fpr_train, tpr_train, th_train = metrics.roc_curve(y_train, train_proba)



plt.figure(figsize=[4, 4])

plt.plot(fpr_test, tpr_test, '')

plt.plot(fpr_train, tpr_train, '')

plt.title('ROC curve')

plt.show()



print('AUC = %6.4f' %metrics.auc(fpr_test, tpr_test))





# # 网格搜索寻参



# In[17]:







import sklearn.tree as tree

from sklearn.model_selection import GridSearchCV

from sklearn import metrics



import sklearn.model_selection as cross_validation

train_data, test_data, train_target, test_target = cross_validation.train_test_split(data,target, test_size=0.4, train_size=0.6 ,random_state=12345)







# In[18]:





param_grid = {

    'criterion':['entropy','gini'],

    'max_depth':[2,3,4],

    'min_samples_split':[2,4,6,8,10] 

}

clf = tree.DecisionTreeClassifier()

clfcv = GridSearchCV(estimator=clf, param_grid=param_grid, 

                   scoring='roc_auc', cv=4)

clfcv.fit(train_data, train_target)



clfcv.best_params_







# In[19]:





clf = tree.DecisionTreeClassifier(criterion='entropy', max_depth=3, min_samples_split=2) 

clf.fit(train_data, train_target) 





train_est = clf.predict(train_data) 

train_est_p=clf.predict_proba(train_data)[:,1]

test_est=clf.predict(test_data)

test_est_p=clf.predict_proba(test_data)[:,1] 

pd.DataFrame({'test_target':test_target,'test_est':test_est,'test_est_p':test_est_p})







# In[20]:





import matplotlib.pyplot as plt

fpr_test, tpr_test, th_test = metrics.roc_curve(test_target, test_est_p)

fpr_train, tpr_train, th_train = metrics.roc_curve(train_target, train_est_p)

plt.figure(figsize=[6,6])

plt.plot(fpr_test, tpr_test)

plt.plot(fpr_train, tpr_train)

plt.title('ROC curve')

plt.show()



print('AUC = %6.4f' %metrics.auc(fpr_test, tpr_test))







# In[21]:





import pydotplus

from IPython.display import Image

import sklearn.tree as tree



dot_data = tree.export_graphviz(

    clf, 

    out_file=None, 

    feature_names=data.columns,

    max_depth=5,

    class_names=['0','1'],

    filled=True

) 

            

graph = pydotplus.graph_from_dot_data(dot_data)  

Image(graph.create_png()) 





# In[22]:







from sklearn.preprocessing import MinMaxScaler

scaler = MinMaxScaler()

scaler.fit(train_data)

scaled_train_data = scaler.transform(train_data)

scaled_test_data = scaler.transform(test_data)





# In[23]:







from sklearn.neural_network import MLPClassifier

param_grid = {

    'hidden_layer_sizes':[(10, ), (15, ), (20, ), (5, 5)],

    'activation':['logistic', 'tanh', 'relu'], 

    'alpha':[0.001, 0.01, 0.1, 0.2, 0.4, 1, 10]

}

mlp = MLPClassifier(max_iter=1000)

gcv = GridSearchCV(estimator=mlp, param_grid=param_grid, 

                   scoring='roc_auc', cv=4, n_jobs=-1)

gcv.fit(scaled_train_data, train_target)



gcv.best_params_



test_est_p=gcv.predict_proba(scaled_test_data)[:,1]

train_est_p=gcv.predict_proba(scaled_train_data)[:,1]





# In[24]:







fpr_test, tpr_test, th_test = metrics.roc_curve(test_target, test_est_p)

fpr_train, tpr_train, th_train = metrics.roc_curve(train_target, train_est_p)

plt.figure(figsize=[4, 4])

plt.plot(fpr_test, tpr_test, '')

plt.plot(fpr_train, tpr_train, '')

plt.title('ROC curve')

plt.show()

print('AUC = %6.4f' %metrics.auc(fpr_test, tpr_test))





# # 变量分箱



# In[25]:





import os

import pandas as pd



data = pd.read_csv('date_data2.csv', skipinitialspace=True)



target = data['Dated']

data = data.ix[:, :4]

data.head()







# In[26]:







import sklearn.tree as tree

from sklearn.model_selection import GridSearchCV

from sklearn import metrics



import sklearn.model_selection as cross_validation

train_data, test_data, train_target, test_target = cross_validation.train_test_split(data,target, test_size=0.4, train_size=0.6 ,random_state=12345) # 划分训练集和测试集





# In[27]:







param_grid = {

    'criterion':['entropy','gini'],

    'max_depth':[2,3,4],

    'min_samples_split':[2,4,6,8,10] 

}

clf = tree.DecisionTreeClassifier()

clfcv = GridSearchCV(estimator=clf, param_grid=param_grid, 

                   scoring='roc_auc', cv=4)

clfcv.fit(train_data, train_target)

#%%

clfcv.best_params_





# In[28]:







clf = tree.DecisionTreeClassifier(criterion='entropy', max_depth=3, min_samples_split=2) 

clf.fit(train_data, train_target) 





train_est = clf.predict(train_data)

train_est_p=clf.predict_proba(train_data)[:,1]

test_est=clf.predict(test_data)

test_est_p=clf.predict_proba(test_data)[:,1] 

pd.DataFrame({'test_target':test_target,'test_est':test_est,'test_est_p':test_est_p})









# In[29]:





import matplotlib.pyplot as plt

fpr_test, tpr_test, th_test = metrics.roc_curve(test_target, test_est_p)

fpr_train, tpr_train, th_train = metrics.roc_curve(train_target, train_est_p)

plt.figure(figsize=[6,6])

plt.plot(fpr_test, tpr_test)

plt.plot(fpr_train, tpr_train)

plt.title('ROC curve')

plt.show()



print('AUC = %6.4f' %metrics.auc(fpr_test, tpr_test))





# In[30]:





import pydotplus

from IPython.display import Image

import sklearn.tree as tree



dot_data = tree.export_graphviz(

    clf, 

    out_file=None, 

    feature_names=data.columns,

    max_depth=5,

    class_names=['0','1'],

    filled=True

) 

            

graph = pydotplus.graph_from_dot_data(dot_data)  

Image(graph.create_png()) 





# In[31]:







train_data, test_data, train_target, test_target = cross_validation.train_test_split(data,target, test_size=0.4, train_size=0.6 ,random_state=12345) 

from sklearn.preprocessing import MinMaxScaler

scaler = MinMaxScaler()

scaler.fit(train_data)

scaled_train_data = scaler.transform(train_data)

scaled_test_data = scaler.transform(test_data)







# In[32]:





from sklearn.neural_network import MLPClassifier

param_grid = {

    'hidden_layer_sizes':[(10, ), (15, ), (20, ), (5, 5)],

    'activation':['logistic', 'tanh', 'relu'], 

    'alpha':[0.001, 0.01, 0.1, 0.2, 0.4, 1, 10]

}



mlp = MLPClassifier(max_iter=1000)

gcv = GridSearchCV(estimator=mlp, param_grid=param_grid, 

                   scoring='roc_auc', cv=4, n_jobs=-1)

gcv.fit(scaled_train_data, train_target)





gcv.best_params_





# In[33]:





test_est_p=gcv.predict_proba(scaled_test_data)[:,1]

train_est_p=gcv.predict_proba(scaled_train_data)[:,1]





# In[34]:





fpr_test, tpr_test, th_test = metrics.roc_curve(test_target, test_est_p)

fpr_train, tpr_train, th_train = metrics.roc_curve(train_target, train_est_p)

plt.figure(figsize=[4, 4])

plt.plot(fpr_test, tpr_test, '')

plt.plot(fpr_train, tpr_train, '')

plt.title('ROC curve')

plt.show()

print('AUC = %6.4f' %metrics.auc(fpr_test, tpr_test))





# In[ ]: