2022第五届“泰迪杯”数据分析技能赛-B题-银行客户忠诚度分析（下）

2022第五届“泰迪杯”数据分析技能赛-B题-银行客户忠诚度分析(上)链接：https://blog.csdn.net/weixin_60200880/article/details/127917207?spm=1001.2014.3001.5502

题目链接：链接:https://pan.baidu.com/s/1x1U-kobvPjNMm8xnvS9Gdg
提取码:7id3

任务4 特征构建

任务5 银行客户长期忠诚度预测建模

任务5.1

任务5.2

比赛总结

good

bad

任务4 特征构建

基于长期数据提取影响客户流失的因素，构建与银行客户长期忠诚度相关的特征，将结果保存到文件“result4.xlsx”中。 (1) 根据表 7，构建新老客户活跃程度的特征，并将结果存于“IsActiveStatus” 列

(2) 根据表 8，构建不同金融资产客户活跃程度的特征，并将结果存于“IsActiveAssetStage”列。

(3) 根据表 9，构建不同金融资产信用卡持有状态的特征，并将结果存于“CrCardAssetStage”列。

import pandas as pd
long_data=pd.read_excel('F:\\泰迪杯B题\\B题：银行客户忠诚度分析赛题数据\\任务4.xlsx')

IsActiveStatus=[]
IsActiveAssetStaget=[]
CrCardAssetStage=[]
for i in range(len(long_data["Age"])):
    IsActiveStatus.append(0)
    IsActiveAssetStaget.append(0)
    CrCardAssetStage.append(0)
for i in range(len(short_data["Age"])):
    if long_data.loc[i,'Status']=='新客户' and long_data.loc[i,'IsActiveMember']==0:
        IsActiveStatus[i]=0
    if long_data.loc[i,'Status']=='稳定客户' and long_data.loc[i,'IsActiveMember']==0:
        IsActiveStatus[i]=1
    if long_data.loc[i,'Status']=='老客户' and long_data.loc[i,'IsActiveMember']==0:
        IsActiveStatus[i]=2
    if long_data.loc[i,'Status']=='新客户' and long_data.loc[i,'IsActiveMember']==1:
        IsActiveStatus[i]=3
    if long_data.loc[i,'Status']=='稳定客户' and long_data.loc[i,'IsActiveMember']==1:
        IsActiveStatus[i]=4
    if long_data.loc[i,'Status']=='老客户' and long_data.loc[i,'IsActiveMember']==1:
        IsActiveStatus[i]=5
    
for i in range(len(short_data["Age"])):
    if long_data.loc[i,'AssetStage']=='低资产' and long_data.loc[i,'IsActiveMember']==0:
        IsActiveAssetStaget[i]=0
    if long_data.loc[i,'AssetStage']=='中下资产' and long_data.loc[i,'IsActiveMember']==0:
        IsActiveAssetStaget[i]=1
    if long_data.loc[i,'AssetStage']=='中上资产' and long_data.loc[i,'IsActiveMember']==0:
        IsActiveAssetStaget[i]=2
    if long_data.loc[i,'AssetStage']=='高资产' and long_data.loc[i,'IsActiveMember']==0:
        IsActiveAssetStaget[i]=3
    if long_data.loc[i,'AssetStage']=='低资产' and long_data.loc[i,'IsActiveMember']==1:
        IsActiveAssetStaget[i]=6
    if long_data.loc[i,'AssetStage']=='中下资产' and long_data.loc[i,'IsActiveMember']==1:
        IsActiveAssetStaget[i]=7
    if long_data.loc[i,'AssetStage']=='中上资产' and long_data.loc[i,'IsActiveMember']==1:
        IsActiveAssetStaget[i]=8
    if long_data.loc[i,'AssetStage']=='高资产' and long_data.loc[i,'IsActiveMember']==1:
        IsActiveAssetStaget[i]=9
    
for i in range(len(short_data["Age"])):
    if long_data.loc[i,'AssetStage']=='低资产' and long_data.loc[i,'HasCrCard']==0:
        CrCardAssetStage[i]=0
    if long_data.loc[i,'AssetStage']=='中下资产' and long_data.loc[i,'HasCrCard']==0:
        CrCardAssetStage[i]=2
    if long_data.loc[i,'AssetStage']=='中上资产' and long_data.loc[i,'HasCrCard']==0:
        CrCardAssetStage[i]=5
    if long_data.loc[i,'AssetStage']=='高资产' and long_data.loc[i,'HasCrCard']==0:
        CrCardAssetStage[i]=5
    if long_data.loc[i,'AssetStage']=='低资产' and long_data.loc[i,'HasCrCard']==1:
        CrCardAssetStage[i]=6
    if long_data.loc[i,'AssetStage']=='中下资产' and long_data.loc[i,'HasCrCard']==1:
        CrCardAssetStage[i]=7
    if long_data.loc[i,'AssetStage']=='中上资产' and long_data.loc[i,'HasCrCard']==1:
        CrCardAssetStage[i]=9
    if long_data.loc[i,'AssetStage']=='高资产' and long_data.loc[i,'HasCrCard']==1:
        CrCardAssetStage[i]=9

data_1={'IsActiveStatus':IsActiveStatus,'IsActiveAssetStaget':IsActiveAssetStaget,'CrCardAssetStage':CrCardAssetStage}
df_1=pd.DataFrame(data_1)
df_1.to_excel('F:\\泰迪杯B题\\B题：银行客户忠诚度分析赛题数据\\result4.xlsx',sheet_name='sheet1',index=None)

任务5 银行客户长期忠诚度预测建模

长期数据存在“Exited”特征分布不均衡、各项数值分布跨度大等现象。体现为：未流失客户量是已流失客户量的 3 倍以上；客户信用资格最大数值达到 25万，而客户活动状态则为 0 和 1 等。考虑上述现象，对银行客户长期忠诚度进行预测。

任务5.1

(1) 选取适当的客户特征，建立客户长期忠诚度预测模型。客户特征可以从客户信用资格、性别、年龄、户龄、金融资产、客户购买产品数量、持有信用卡状态、活动状态和个人年收入等指标中直接选取，也可以参照任务 4 构建。在报告中给出特征选取的依据、建立预测模型的思路和过程。 (2) 使用混淆矩阵、F1 Score 等方法对预测模型进行评估，在报告中给出评估的方法和结果。1.导入库

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import missingno as ms
import plotly.express as px
import plotly.graph_objs as go
import plotly.figure_factory as ff
from plotly.subplots import make_subplots
import plotly.offline as pyo
pyo.init_notebook_mode()
sns.set_style('darkgrid')
plt.style.use('fivethirtyeight')
%matplotlib inline

from sklearn.decomposition import PCA
from imblearn.over_sampling import SMOTE
from sklearn.model_selection import train_test_split,cross_val_score
from sklearn.ensemble import RandomForestClassifier,AdaBoostClassifier
from sklearn.neighbors import KNeighborsClassifier
from sklearn.metrics import accuracy_score, confusion_matrix, classification_report
from sklearn.svm import SVC
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import f1_score as f1
from sklearn.metrics import confusion_matrix

import eli5
from eli5.sklearn import PermutationImportance
import shap
plt.rc('figure',figsize=(18,9))

import warnings
warnings.filterwarnings("ignore")
pd.set_option('display.max_columns', 26)
long_data=pd.read_excel('F:\\泰迪杯B题\\B题：银行客户忠诚度分析赛题数据\\result1_2.xlsx')

2.建模

x=long_data.drop("Exited",axis=1)
y=long_data["Exited"]
# 随机打乱数据
from sklearn.utils import shuffle
data_train=shuffle(long_data)
x_train,x_test,y_train,y_test=train_test_split(x,y,test_size=0.30,random_state=0)

3.定义建模函数

def create_model(model):
    #模型训练
    model.fit(x_train,y_train)
    #模型预测
    y_pred=model.predict(x_test)
    #准确率acc
    acc=accuracy_score(y_test,y_pred)
    #混淆矩阵
    cm=confusion_matrix(y_test,y_pred)
    #分类报告
    cr=classification_report(y_test,y_pred)
    
    print(f"Test Accuracy of {model}:{acc}")
    print(f"Confusion Matrix of {model}:\n{cm}")
    print(f"Classification Report of {model}:\n{cr}")

4.十种模型

4.1 KNN

from sklearn.neighbors import KNeighborsClassifier
knn=KNeighborsClassifier()
create_model(knn)

---------------------代码分割线-----------------------------

Test Accuracy of KNeighborsClassifier():0.8041199855439104
Confusion Matrix of KNeighborsClassifier():
[[2106  112]
 [ 430  119]]
Classification Report of KNeighborsClassifier():
              precision    recall  f1-score   support

           0       0.83      0.95      0.89      2218
           1       0.52      0.22      0.31       549

    accuracy                           0.80      2767
   macro avg       0.67      0.58      0.60      2767
weighted avg       0.77      0.80      0.77      2767

4.2 DecisionTreeClassifier

from sklearn.tree import DecisionTreeClassifier
dt=DecisionTreeClassifier()
create_model(dt)

-----------------------代码分割线------------------------

Test Accuracy of DecisionTreeClassifier():0.8062883989880737
Confusion Matrix of DecisionTreeClassifier():
[[1919  299]
 [ 237  312]]
Classification Report of DecisionTreeClassifier():
              precision    recall  f1-score   support

           0       0.89      0.87      0.88      2218
           1       0.51      0.57      0.54       549

    accuracy                           0.81      2767
   macro avg       0.70      0.72      0.71      2767
weighted avg       0.81      0.81      0.81      2767

4.3 RandomForestClassifier

from sklearn.ensemble import RandomForestClassifier
rd_clf=RandomForestClassifier(criterion='entropy'
                             ,max_depth=11
                             ,max_features='auto'
                             ,min_samples_leaf=2
                             ,min_samples_split=3
                             ,n_estimators=130)
create_model(rd_clf)

------------------------------------代码分割线--------------------------------------

Test Accuracy of RandomForestClassifier(criterion='entropy', max_depth=11, max_features='auto',
                       min_samples_leaf=2, min_samples_split=3,
                       n_estimators=130):0.8782074448861583
Confusion Matrix of RandomForestClassifier(criterion='entropy', max_depth=11, max_features='auto',
                       min_samples_leaf=2, min_samples_split=3,
                       n_estimators=130):
[[2165   53]
 [ 284  265]]
Classification Report of RandomForestClassifier(criterion='entropy', max_depth=11, max_features='auto',
                       min_samples_leaf=2, min_samples_split=3,
                       n_estimators=130):
              precision    recall  f1-score   support

           0       0.88      0.98      0.93      2218
           1       0.83      0.48      0.61       549

    accuracy                           0.88      2767
   macro avg       0.86      0.73      0.77      2767
weighted avg       0.87      0.88      0.86      2767

4.4 AdaBoostClassifier

from sklearn.ensemble import AdaBoostClassifier
ada=AdaBoostClassifier(base_estimator=dt)
create_model(ada)

------------------------------------------代码分割线-------------------------------------

Test Accuracy of AdaBoostClassifier(base_estimator=DecisionTreeClassifier()):0.8091796169136248
Confusion Matrix of AdaBoostClassifier(base_estimator=DecisionTreeClassifier()):
[[1927  291]
 [ 237  312]]
Classification Report of AdaBoostClassifier(base_estimator=DecisionTreeClassifier()):
              precision    recall  f1-score   support

           0       0.89      0.87      0.88      2218
           1       0.52      0.57      0.54       549

    accuracy                           0.81      2767
   macro avg       0.70      0.72      0.71      2767
weighted avg       0.82      0.81      0.81      2767

4.5 GradientBoostingClassifier

from sklearn.ensemble import GradientBoostingClassifier
gb=GradientBoostingClassifier()
create_model(gb)

----------------------------------代码分割线----------------------------------------

Test Accuracy of GradientBoostingClassifier():0.878930249367546
Confusion Matrix of GradientBoostingClassifier():
[[2155   63]
 [ 272  277]]
Classification Report of GradientBoostingClassifier():
              precision    recall  f1-score   support

           0       0.89      0.97      0.93      2218
           1       0.81      0.50      0.62       549

    accuracy                           0.88      2767
   macro avg       0.85      0.74      0.78      2767
weighted avg       0.87      0.88      0.87      2767

4.6 XGBClassifier

from xgboost import XGBClassifier
xgb=XGBClassifier(objective='binary:logistic'
                 ,learning_rate=0.5
                 ,max_depth=5
                 ,n_estimators=150)
create_model(xgb)

----------------------------------------------代码分割线------------------------------------

Test Accuracy of XGBClassifier(base_score=0.5, booster='gbtree', callbacks=None,
              colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,
              early_stopping_rounds=None, enable_categorical=False,
              eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
              importance_type=None, interaction_constraints='',
              learning_rate=0.5, max_bin=256, max_cat_to_onehot=4,
              max_delta_step=0, max_depth=5, max_leaves=0, min_child_weight=1,
              missing=nan, monotone_constraints='()', n_estimators=150,
              n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=0,
              reg_alpha=0, reg_lambda=1, ...):0.8641127574990964
Confusion Matrix of XGBClassifier(base_score=0.5, booster='gbtree', callbacks=None,
              colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,
              early_stopping_rounds=None, enable_categorical=False,
              eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
              importance_type=None, interaction_constraints='',
              learning_rate=0.5, max_bin=256, max_cat_to_onehot=4,
              max_delta_step=0, max_depth=5, max_leaves=0, min_child_weight=1,
              missing=nan, monotone_constraints='()', n_estimators=150,
              n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=0,
              reg_alpha=0, reg_lambda=1, ...):
[[2106  112]
 [ 264  285]]
Classification Report of XGBClassifier(base_score=0.5, booster='gbtree', callbacks=None,
              colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,
              early_stopping_rounds=None, enable_categorical=False,
              eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
              importance_type=None, interaction_constraints='',
              learning_rate=0.5, max_bin=256, max_cat_to_onehot=4,
              max_delta_step=0, max_depth=5, max_leaves=0, min_child_weight=1,
              missing=nan, monotone_constraints='()', n_estimators=150,
              n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=0,
              reg_alpha=0, reg_lambda=1, ...):
              precision    recall  f1-score   support

           0       0.89      0.95      0.92      2218
           1       0.72      0.52      0.60       549

    accuracy                           0.86      2767
   macro avg       0.80      0.73      0.76      2767
weighted avg       0.85      0.86      0.86      2767

4.7 CatBoostClassifier

from catboost import CatBoostClassifier
cab=CatBoostClassifier(iterations=10)
create_model(cab)

--------------------------------------代码分割线--------------------------------------------

Learning rate set to 0.5
0:	learn: 0.4772704	total: 142ms	remaining: 1.27s
1:	learn: 0.4010936	total: 143ms	remaining: 570ms
2:	learn: 0.3680817	total: 144ms	remaining: 336ms
3:	learn: 0.3486839	total: 145ms	remaining: 218ms
4:	learn: 0.3360775	total: 147ms	remaining: 147ms
5:	learn: 0.3280172	total: 148ms	remaining: 98.7ms
6:	learn: 0.3221241	total: 150ms	remaining: 64.1ms
7:	learn: 0.3176367	total: 151ms	remaining: 37.7ms
8:	learn: 0.3148666	total: 152ms	remaining: 16.9ms
9:	learn: 0.3115188	total: 153ms	remaining: 0us
Test Accuracy of <catboost.core.CatBoostClassifier object at 0x000002D7908D6880>:0.8760390314419949
Confusion Matrix of <catboost.core.CatBoostClassifier object at 0x000002D7908D6880>:
[[2148   70]
 [ 273  276]]
Classification Report of <catboost.core.CatBoostClassifier object at 0x000002D7908D6880>:
              precision    recall  f1-score   support

           0       0.89      0.97      0.93      2218
           1       0.80      0.50      0.62       549

    accuracy                           0.88      2767
   macro avg       0.84      0.74      0.77      2767
weighted avg       0.87      0.88      0.86      2767

4.8 ExtraTreesClassifier

from sklearn.ensemble import ExtraTreesClassifier
etc=ExtraTreesClassifier()
create_model(etc)

-------------------------------代码分割线--------------------------------------

Test Accuracy of ExtraTreesClassifier():0.8695337911095049
Confusion Matrix of ExtraTreesClassifier():
[[2158   60]
 [ 301  248]]
Classification Report of ExtraTreesClassifier():
              precision    recall  f1-score   support

           0       0.88      0.97      0.92      2218
           1       0.81      0.45      0.58       549

    accuracy                           0.87      2767
   macro avg       0.84      0.71      0.75      2767
weighted avg       0.86      0.87      0.85      2767

4.9 LGBMClassifier

from lightgbm import LGBMClassifier
lgbm=LGBMClassifier(learning_rate=0.1)
create_model(lgbm)

-----------------------------代码分割线------------------------------------------

Test Accuracy of LGBMClassifier():0.8764004336826888
Confusion Matrix of LGBMClassifier():
[[2143   75]
 [ 267  282]]
Classification Report of LGBMClassifier():
              precision    recall  f1-score   support

           0       0.89      0.97      0.93      2218
           1       0.79      0.51      0.62       549

    accuracy                           0.88      2767
   macro avg       0.84      0.74      0.77      2767
weighted avg       0.87      0.88      0.87      2767

4.10 svm

from sklearn import svm
wclf=svm.SVC(kernel='rbf',class_weight={1:3})
create_model(wclf)

--------------------------------代码分割线------------------------------------

Test Accuracy of SVC(class_weight={1: 3}):0.8015901698590532
Confusion Matrix of SVC(class_weight={1: 3}):
[[2218    0]
 [ 549    0]]
Classification Report of SVC(class_weight={1: 3}):
              precision    recall  f1-score   support

           0       0.80      1.00      0.89      2218
           1       0.00      0.00      0.00       549

    accuracy                           0.80      2767
   macro avg       0.40      0.50      0.44      2767
weighted avg       0.64      0.80      0.71      2767

5.绘制模型评分直方图

models=pd.DataFrame({"model":["SVM","KNN","Decision Tree","Random Forest","Ada Boost","Gradient Boosting","XgBoost","catBoost","Extra Trees","LGBM"],
                    "acc":[0.802,0.804,0.806,0.878,0.809,0.879,0.864,0.876,0.869,0.876]})
models=models.sort_values("acc",ascending=True)
px.bar(models
      ,x="acc"
      ,y="model"
      ,text="acc"
      ,color='acc'
      ,template='plotly_dark'
      ,title='Models Comparision')

6.模型可解释性

在这里选择Gradient Boost（gb）同时使用shap库来进行解释

6.1 shap值计算

# 模型可解释性,选择Extra Tree(etc)
explainer=shap.TreeExplainer(gb)
# 在explainer中传入特征值的数据，计算shap值
shap_values=explainer.shap_values(x_test)
shap_values

-------------------------------------------------------------------

array([[-1.62093641e-01, -2.90161088e-02,  1.17664011e-01, ...,
        -1.07458257e-05, -3.58140368e-01,  1.20403416e-02],
       [-1.51243735e-01, -4.63661455e-02,  1.47073454e-01, ...,
        -1.07458257e-05,  4.34854369e-01, -1.76490066e-01],
       [-1.93332535e-01, -6.02414362e-02, -1.25060216e-01, ...,
         1.07458257e-05, -2.47356490e-01, -3.77205325e-02],
       ...,
       [-6.51192238e-02, -3.84451253e-02,  1.61559034e-01, ...,
         4.56697590e-05, -2.59919373e-01,  2.10393113e-02],
       [-1.50441872e-01,  1.86815435e-03, -1.54295830e-01, ...,
        -1.07458257e-05,  3.89491535e-01, -4.21581673e-02],
       [-1.76659921e-01, -1.06582926e-01, -1.20482271e-01, ...,
         1.07458257e-05, -2.99516880e-01,  4.03566576e-02]])

6.2 Feature importance

shap.summary_plot(shap_values,x_test,plot_type="bar")

shap.summary_plot(shap_values,x_test)

横坐标为 shap_value，即影响的权重，而非特征的具体指，特征值的大小对结果的影响通过颜色表示，红色值大，蓝色值小，紫色邻近均值。因此，区域分布越宽说明它的影响力越大。

任务5.2

(3) 对“long-customer-test.csv”测试数据进行预测，将全部预测结果以表 10 形式保存为文件“result5.xlsx”，其中 0 表示客户没有流失，1 表示客户流失。并将表 11 中的 5 个客户 ID 的预测结果在报告中列出。

fact_data=pd.read_csv('F:\泰迪杯B题\B题：银行客户忠诚度分析赛题数据\B题：银行客户忠诚度分析赛题数据\long-customer-test.csv')
pred=gb.predict(fact_data)
CustomerId=[]
for i in range(len(fact_data["CustomerId"])):
    CustomerId.append(0)
    
for i in range(len(fact_data["CustomerId"])):
        CustomerId[i]=fact_data.loc[i,"CustomerId"]

pred

--------------------------------------------------------------------------------------

array([0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0,
       1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0,
       0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       1, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0,
       0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0,
       0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0,
       1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0,
       0, 0, 1, 1, 0, 0, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0,
       0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 1, 0, 0, 0, 0, 1, 0, 1, 1, 0, 0, 0,
       1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1,
       1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1,
       0, 1, 0, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1,
       0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0,
       1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0,
       0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0,
       0, 0, 1, 0, 1, 1, 1, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0,
       1, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 1, 1, 0, 1,
       0, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0,
       0, 0, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0,
       1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1,
       0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 1, 0,
       0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0,
       1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1,
       0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0,
       0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 1, 0, 0, 1, 1, 0,
       0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0,
       0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0,
       0, 1, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0,
       0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0,
       0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0,
       0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0,
       0, 0, 1, 0, 0, 0, 0, 1, 0, 0], dtype=int64)

写入文件

data_1={'CustomerId':CustomerId,'Exited':pred}
df_1=pd.DataFrame(data_1)
df_1.to_excel('F:\\泰迪杯B题\\B题：银行客户忠诚度分析赛题数据\\result5.xlsx',sheet_name='sheet1',index=None)

部分预测结果如下：

Customerld	Exited
15779131	0
15674442	0
15719508	1
15730076	1
15792228	1

至此，整个题目基本完成！

比赛总结

首先，这是我第一次单独一个人参加并完成比赛，这个比赛组队规则是1-3人组队，我没有组队的原因其实有很多（这里就不说了……）。接下来说说这次比赛好的方面和一些不足的地方吧。

good

好的方面当然是题目能做完且按时提交啦，因为B题比赛时间是周日，A题比赛时间是周六，然后第一天A题的题目确实是有点不按套路出牌，我本来给自己设定的时间是下午6点就开始写论文的，结果到6点题目只做了一半不到(o(╥﹏╥)o)，最后还是放弃做题开始写论文，整个过程都很不尽人意。B题能做完主要是算法部分有提前准备算法和代码，所以整个过程感觉还是良好的，也积累了不少经验。

bad

1.最大最大的问题：虽然题目有做完，但是论文质量不高(可能是第一次写论文的原因吧，之前都是队友写的)，最大的问题就是论文忘了加标题，论文的排版、说明、过程叙述也都不是很好。

2.比赛过程也反映了自己对代码不够熟练，整道题基本每解决一个小任务都要csdn

3.比赛状态也不是很好，睡眠不足+能量不足(因为单独参赛赶时间，直到下午两点辅导员催做核酸顺路在饭堂买了点吃的)

总之，能在比赛中找到自己的不足还是值得的，争取下次能做得更好吧！

接下来该准备期末考了(⊙﹏⊙)

文章出处登录后可见！

已经登录？立即刷新

2022第五届“泰迪杯”数据分析技能赛-B题-银行客户忠诚度分析（下）

任务4 特征构建

任务5 银行客户长期忠诚度预测建模

任务5.1

任务5.2

比赛总结

good

bad

相关推荐