第二届翼支付杯大数据建模大赛-信用风险用户识别Baseline 线上0.65+稳进复赛

最新推荐文章于 2024-08-02 18:10:59 发布
最新推荐文章于 2024-08-02 18:10:59 发布
阅读量1.7k 收藏 7
点赞数 1
文章标签： 机器学习
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/weixin_44015027/article/details/107628055
版权

导入包

# coding: utf-8
#导入包
import pandas as pd
import os
import gc
import lightgbm as lgb
import xgboost as xgb
from catboost import CatBoostRegressor
from sklearn.linear_model import SGDRegressor, LinearRegression, Ridge
from sklearn.preprocessing import MinMaxScaler
import math
import numpy as np
from tqdm import tqdm
from sklearn.model_selection import StratifiedKFold, KFold
from sklearn.metrics import mean_absolute_error, mean_squared_error
import matplotlib.pyplot as plt
import time
import warnings
warnings.filterwarnings('ignore')

读取训练集数据

#读取训练集
train_base = pd.read_csv('train/train_base.csv')
train_op = pd.read_csv('train/train_op.csv')
train_trans = pd.read_csv('train/train_trans.csv')
train_label = pd.read_csv('train/train_label.csv')

读取测试集数据

#读取测试集
test_base = pd.read_csv('test/test_a_base.csv')
test_op = pd.read_csv('test/test_a_op.csv')
test_trans = pd.read_csv('test/test_a_trans.csv')

我们发现训练集和测试集中sevice3_level字段缺失较多，因此删除此列。然后删除训练集中的sex列以及balance_avg列为空的行，得到不含缺失值的训练集。
对于测试集，其他列缺失值很少，直接用上一行的数据补全。

#去除训练集缺失值
train_df = train_df[train_df['sex'].notnull()]
train_df = train_df[train_df['balance_avg'].notnull()]
train_df.drop('service3_level',axis=1,inplace=True)
train_df.head()

test_df.drop('service3_level',axis=1,inplace=True)
test_df.fillna(method='bfill',inplace=True)
test_df.info()

对于base数据集中的类别编码字段，采用label encode，对于有明显顺序关系和类别数目较少的特征，采用label编码；对于类别较多且没有顺序关系的特征，采取暂时舍弃特征的方式。

#label编码
train_df['sex'].loc[train_df['sex']=='category 1'] = 1
train_df['sex'].loc[train_df['sex']=='category 0'] = 0

train_df['level'].loc[train_df['level']=='category 0'] = 0
train_df['level'].loc[train_df['level']=='category 1'] = 1
train_df['level'].loc[train_df['level']=='category 2'] = 2

train_df['verified'].loc[train_df['verified']=='category 1'] = 1
train_df['verified'].loc[train_df['verified']=='category 0'] = 0

train_df['agreement1'].loc[train_df['agreement1']=='category 1'] = 1
train_df['agreement1'].loc[train_df['agreement1']=='category 0'] = 0

train_df['agreement2'].loc[train_df['agreement2']=='category 1'] = 1
train_df['agreement2'].loc[train_df['agreement2']=='category 0'] = 0

train_df['agreement3'].loc[train_df['agreement3']=='category 1'] = 1
train_df['agreement3'].loc[train_df['agreement3']=='category 0'] = 0

train_df['agreement4'].loc[train_df['agreement4']=='category 1'] = 1
train_df['agreement4'].loc[train_df['agreement4']=='category 0'] = 0

train_df['balance'].loc[train_df['balance']=='level 0'] = 0
train_df['balance'].loc[train_df['balance']=='level 1'] = 1
train_df['balance'].loc[train_df['balance']=='level 2'] = 2
train_df['balance'].loc[train_df['balance']=='level 3'] = 3
train_df['balance'].loc[train_df['balance']=='level 4'] = 4
train_df['balance'].loc[train_df['balance']=='level 5'] = 5
train_df['balance'].loc[train_df['balance']=='level 6'] = 6
train_df['balance'].loc[train_df['balance']=='level 7'] = 7
train_df['balance'].loc[train_df['balance']=='level 8'] = 8
train_df['balance'].loc[train_df['balance']=='level 9'] = 9
train_df['balance'].loc[train_df['balance']=='level 10'] = 10
train_df['balance'].loc[train_df['balance']=='level 11'] = 11
train_df['balance'].loc[train_df['balance']=='level 12'] = 12
train_df['balance'].loc[train_df['balance']=='level 13'] = 13
train_df['balance'].loc[train_df['balance']=='level 14'] = 14
train_df['balance'].loc[train_df['balance']=='level 15'] = 15
train_df['balance'].loc[train_df['balance']=='level 16'] = 16
train_df['balance'].loc[train_df['balance']=='level 17'] = 17
train_df['balance'].loc[train_df['balance']=='level 18'] = 18
train_df['balance'].loc[train_df['balance']=='level 19'] = 19
train_df['balance'].loc[train_df['balance']=='level 20'] = 20
train_df['balance'].loc[train_df['balance']=='level 21'] = 21

train_df['balance_avg'].loc[train_df['balance_avg']=='level 0'] = 0
train_df['balance_avg'].loc[train_df['balance_avg']=='level 1'] = 1
train_df['balance_avg'].loc[train_df['balance_avg']=='level 2'] = 2
train_df['balance_avg'].loc[train_df['balance_avg']=='level 3'] = 3
train_df['balance_avg'].loc[train_df['balance_avg']=='level 4'] = 4
train_df['balance_avg'].loc[train_df['balance_avg']=='level 5'] = 5
train_df['balance_avg'].loc[train_df['balance_avg']=='level 6'] = 6
train_df['balance_avg'].loc[train_df['balance_avg']=='level 7'] = 7
train_df['balance_avg'].loc[train_df['balance_avg']=='level 8'] = 8
train_df['balance_avg'].loc[train_df['balance_avg']=='level 9'] = 9
train_df['balance_avg'].loc[train_df['balance_avg']=='level 10'] = 10
train_df['balance_avg'].loc[train_df['balance_avg']=='level 11'] = 11
train_df['balance_avg'].loc[train_df['balance_avg']=='level 12'] = 12
train_df['balance_avg'].loc[train_df['balance_avg']=='level 13'] = 13
train_df['balance_avg'].loc[train_df['balance_avg']=='level 14'] = 14
train_df['balance_avg'].loc[train_df['balance_avg']=='level 15'] = 15
train_df['balance_avg'].loc[train_df['balance_avg']=='level 16'] = 16
train_df['balance_avg'].loc[train_df['balance_avg']=='level 17'] = 17
train_df['balance_avg'].loc[train_df['balance_avg']=='level 18'] = 18
train_df['balance_avg'].loc[train_df['balance_avg']=='level 19'] = 19
train_df['balance_avg'].loc[train_df['balance_avg']=='level 20'] = 20
train_df['balance_avg'].loc[train_df['balance_avg']=='level 21'] = 21

train_df['balance1'].loc[train_df['balance1']=='level 0'] = 0
train_df['balance1'].loc[train_df['balance1']=='level 1'] = 1
train_df['balance1'].loc[train_df['balance1']=='level 2'] = 2
train_df['balance1'].loc[train_df['balance1']=='level 3'] = 3
train_df['balance1'].loc[train_df['balance1']=='level 4'] = 4
train_df['balance1'].loc[train_df['balance1']=='level 5'] = 5
train_df['balance1'].loc[train_df['balance1']=='level 6'] = 6
train_df['balance1'].loc[train_df['balance1']=='level 7'] = 7
train_df['balance1'].loc[train_df['balance1']=='level 8'] = 8
train_df['balance1'].loc[train_df['balance1']=='level 9'] = 9
train_df['balance1'].loc[train_df['balance1']=='level 10'] = 10
train_df['balance1'].loc[train_df['balance1']=='level 11'] = 11
train_df['balance1'].loc[train_df['balance1']=='level 12'] = 12
train_df['balance1'].loc[train_df['balance1']=='level 13'] = 13
train_df['balance1'].loc[train_df['balance1']=='level 14'] = 14
train_df['balance1'].loc[train_df['balance1']=='level 15'] = 15
train_df['balance1'].loc[train_df['balance1']=='level 16'] = 16
train_df['balance1'].loc[train_df['balance1']=='level 17'] = 17
train_df['balance1'].loc[train_df['balance1']=='level 18'] = 18
train_df['balance1'].loc[train_df['balance1']=='level 19'] = 19
train_df['balance1'].loc[train_df['balance1']=='level 20'] = 20
train_df['balance1'].loc[train_df['balance1']=='level 21'] = 21

train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 0'] = 0
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 1'] = 1
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 2'] = 2
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 3'] = 3
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 4'] = 4
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 5'] = 5
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 6'] = 6
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 7'] = 7
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 8'] = 8
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 9'] = 9
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 10'] = 10
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 11'] = 11
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 12'] = 12
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 13'] = 13
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 14'] = 14
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 15'] = 15
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 16'] = 16
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 17'] = 17
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 18'] = 18
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 19'] = 19
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 20'] = 20
train_df['balance1_avg'].loc[train_df['balance1_avg']=='level 21'] = 21

train_df['balance2'].loc[train_df['balance2']=='level 0'] = 0
train_df['balance2'].loc[train_df['balance2']=='level 1'] = 1
train_df['balance2'].loc[train_df['balance2']=='level 2'] = 2
train_df['balance2'].loc[train_df['balance2']=='level 3'] = 3
train_df['balance2'].loc[train_df['balance2']=='level 4'] = 4
train_df['balance2'].loc[train_df['balance2']=='level 5'] = 5
train_df['balance2'].loc[train_df['balance2']=='level 6'] = 6
train_df['balance2'].loc[train_df['balance2']=='level 7'] = 7
train_df['balance2'].loc[train_df['balance2']=='level 8'] = 8
train_df['balance2'].loc[train_df['balance2']=='level 9'] = 9
train_df['balance2'].loc[train_df['balance2']=='level 10'] = 10
train_df['balance2'].loc[train_df['balance2']=='level 11'] = 11
train_df['balance2'].loc[train_df['balance2']=='level 12'] = 12
train_df['balance2'].loc[train_df['balance2']=='level 13'] = 13
train_df['balance2'].loc[train_df['balance2']=='level 14'] = 14
train_df['balance2'].loc[train_df['balance2']=='level 15'] = 15
train_df['balance2'].loc[train_df['balance2']=='level 16'] = 16
train_df['balance2'].loc[train_df['balance2']=='level 17'] = 17
train_df['balance2'].loc[train_df['balance2']=='level 18'] = 18
train_df['balance2'].loc[train_df['balance2']=='level 19'] = 19
train_df['balance2'].loc[train_df['balance2']=='level 20'] = 20
train_df['balance2'].loc[train_df['balance2']=='level 21'] = 21

train_df['balance2_avg'].loc[train_df['balance2_avg']=='level 0'] = 0
train_df['balance2_avg'].loc[train_df['balance2_avg']=='level 1'] = 1
train_df['balance2_avg'].loc[train_df['balance2_avg']=='level 2'] = 2
train_df['balance2_avg'].loc[train_df['balance2_avg']=='level 3'] = 3
train_df['balance2_avg'].loc[train_df['balance2_avg']=='level 4'] = 4
train_df['balance2_avg'].loc[train_df['balance2_avg']=='level 5'] = 5

train_df['service3'].loc[train_df['service3']=='category 1'] = 1
train_df['service3'].loc[train_df['service3']=='category 0'] = 0

train_df['product1_amount'].loc[train_df['product1_amount']=='level 1'] = 1
train_df['product1_amount'].loc[train_df['product1_amount']=='level 2'] = 2
train_df['product1_amount'].loc[train_df['product1_amount']=='level 3'] = 3
train_df['product1_amount'].loc[train_df['product1_amount']=='level 4'] = 4
train_df['product1_amount'].loc[train_df['product1_amount']=='level 5'] = 5
train_df['product1_amount'].loc[train_df['product1_amount']=='level 6'] = 6
train_df['product1_amount'].loc[train_df['product1_amount']=='level 7'] = 7

train_df['product2_amount'].loc[train_df['product2_amount']=='level 1'] = 1
train_df['product2_amount'].loc[train_df['product2_amount']=='level 2'] = 2
train_df['product2_amount'].loc[train_df['product2_amount']=='level 3'] = 3
train_df['product2_amount'].loc[train_df['product2_amount']=='level 4'] = 4
train_df['product2_amount'].loc[train_df['product2_amount']=='level 5'] = 5
train_df['product2_amount'].loc[train_df['product2_amount']=='level 6'] = 6
train_df['product2_amount'].loc[train_df['product2_amount']=='level 7'] = 7
train_df['product2_amount'].loc[train_df['product2_amount']=='level 8'] = 8
train_df['product2_amount'].loc[train_df['product2_amount']=='level 9'] = 9
train_df['product2_amount'].loc[train_df['product2_amount']=='level 10'] = 10
train_df['product2_amount'].loc[train_df['product2_amount']=='level 11'] = 11
train_df['product2_amount'].loc[train_df['product2_amount']=='level 12'] = 12
train_df['product2_amount'].loc[train_df['product2_amount']=='level 13'] = 13
train_df['product2_amount'].loc[train_df['product2_amount']=='level 14'] = 14
train_df['product2_amount'].loc[train_df['product2_amount']=='level 15'] = 15
train_df['product2_amount'].loc[train_df['product2_amount']=='level 16'] = 16
train_df['product2_amount'].loc[train_df['product2_amount']=='level 17'] = 17
train_df['product2_amount'].loc[train_df['product2_amount']=='level 18'] = 18
train_df['product2_amount'].loc[train_df['product2_amount']=='level 19'] = 19
train_df['product2_amount'].loc[train_df['product2_amount']=='level 20'] = 20
train_df['product2_amount'].loc[train_df['product2_amount']=='level 21'] = 21

train_df['product3_amount'].loc[train_df['product3_amount']=='level 1'] = 1
train_df['product3_amount'].loc[train_df['product3_amount']=='level 2'] = 2
train_df['product3_amount'].loc[train_df['product3_amount']=='level 3'] = 3

train_df['product4_amount'].loc[train_df['product4_amount']=='level 0'] = 0
train_df['product4_amount'].loc[train_df['product4_amount']=='level 1'] = 1

train_df['product5_amount'].loc[train_df['product5_amount']=='level 0'] = 0
train_df['product5_amount'].loc[train_df['product5_amount']=='level 1'] = 1

train_df['product6_amount'].loc[train_df['product6_amount']=='level 1'] = 1
train_df['product6_amount'].loc[train_df['product6_amount']=='level 2'] = 2
train_df['product6_amount'].loc[train_df['product6_amount']=='level 3'] = 3
train_df['product6_amount'].loc[train_df['product6_amount']=='level 4'] = 4
train_df['product6_amount'].loc[train_df['product6_amount']=='level 5'] = 5
train_df['product6_amount'].loc[train_df['product6_amount']=='level 6'] = 6
train_df['product6_amount'].loc[train_df['product6_amount']=='level 7'] = 7
train_df['product6_amount'].loc[train_df['product6_amount']=='level 8'] = 8
train_df['product6_amount'].loc[train_df['product6_amount']=='level 9'] = 9
train_df['product6_amount'].loc[train_df['product6_amount']=='level 10'] = 10
train_df['product6_amount'].loc[train_df['product6_amount']=='level 11'] = 11
train_df['product6_amount'].loc[train_df['product6_amount']=='level 12'] = 12
train_df['product6_amount'].loc[train_df['product6_amount']=='level 13'] = 13
train_df['product6_amount'].loc[train_df['product6_amount']=='level 14'] = 14
train_df['product6_amount'].loc[train_df['product6_amount']=='level 15'] = 15
train_df['product6_amount'].loc[train_df['product6_amount']=='level 16'] = 16
train_df['product6_amount'].loc[train_df['product6_amount']=='level 17'] = 17
train_df['product6_amount'].loc[train_df['product6_amount']=='level 18'] = 18
train_df['product6_amount'].loc[train_df['product6_amount']=='level 19'] = 19
train_df['product6_amount'].loc[train_df['product6_amount']=='level 20'] = 20
train_df['product6_amount'].loc[train_df['product6_amount']=='level 21'] = 21


test_df['sex'].loc[test_df['sex']=='category 1'] = 1
test_df['sex'].loc[test_df['sex']=='category 0'] = 0

test_df['level'].loc[test_df['level']=='category 0'] = 0
test_df['level'].loc[test_df['level']=='category 1'] = 1
test_df['level'].loc[test_df['level']=='category 2'] = 2

test_df['verified'].loc[test_df['verified']=='category 1'] = 1
test_df['verified'].loc[test_df['verified']=='category 0'] = 0

test_df['agreement1'].loc[test_df['agreement1']=='category 1'] = 1
test_df['agreement1'].loc[test_df['agreement1']=='category 0'] = 0

test_df['agreement2'].loc[test_df['agreement2']=='category 1'] = 1
test_df['agreement2'].loc[test_df['agreement2']=='category 0'] = 0

test_df['agreement3'].loc[test_df['agreement3']=='category 1'] = 1
test_df['agreement3'].loc[test_df['agreement3']=='category 0'] = 0

test_df['agreement4'].loc[test_df['agreement4']=='category 1'] = 1
test_df['agreement4'].loc[test_df['agreement4']=='category 0'] = 0

test_df['balance'].loc[test_df['balance']=='level 0'] = 0
test_df['balance'].loc[test_df['balance']=='level 1'] = 1
test_df['balance'].loc[test_df['balance']=='level 2'] = 2
test_df['balance'].loc[test_df['balance']=='level 3'] = 3
test_df['balance'].loc[test_df['balance']=='level 4'] = 4
test_df['balance'].loc[test_df['balance']=='level 5'] = 5
test_df['balance'].loc[test_df['balance']=='level 6'] = 6
test_df['balance'].loc[test_df['balance']=='level 7'] = 7
test_df['balance'].loc[test_df['balance']=='level 8'] = 8
test_df['balance'].loc[test_df['balance']=='level 9'] = 9
test_df['balance'].loc[test_df['balance']=='level 10'] = 10
test_df['balance'].loc[test_df['balance']=='level 11'] = 11
test_df['balance'].loc[test_df['balance']=='level 12'] = 12
test_df['balance'].loc[test_df['balance']=='level 13'] = 13
test_df['balance'].loc[test_df['balance']=='level 14'] = 14
test_df['balance'].loc[test_df['balance']=='level 15'] = 15
test_df['balance'].loc[test_df['balance']=='level 16'] = 16
test_df['balance'].loc[test_df['balance']=='level 17'] = 17
test_df['balance'].loc[test_df['balance']=='level 18'] = 18
test_df['balance'].loc[test_df['balance']=='level 19'] = 19
test_df['balance'].loc[test_df['balance']=='level 20'] = 20
test_df['balance'].loc[test_df['balance']=='level 21'] = 21

test_df['balance_avg'].loc[test_df['balance_avg']=='level 0'] = 0
test_df['balance_avg'].loc[test_df['balance_avg']=='level 1'] = 1
test_df['balance_avg'].loc[test_df['balance_avg']=='level 2'] = 2
test_df['balance_avg'].loc[test_df['balance_avg']=='level 3'] = 3
test_df['balance_avg'].loc[test_df['balance_avg']=='level 4'] = 4
test_df['balance_avg'].loc[test_df['balance_avg']=='level 5'] = 5
test_df['balance_avg'].loc[test_df['balance_avg']=='level 6'] = 6
test_df['balance_avg'].loc[test_df['balance_avg']=='level 7'] = 7
test_df['balance_avg'].loc[test_df['balance_avg']=='level 8'] = 8
test_df['balance_avg'].loc[test_df['balance_avg']=='level 9'] = 9
test_df['balance_avg'].loc[test_df['balance_avg']=='level 10'] = 10
test_df['balance_avg'].loc[test_df['balance_avg']=='level 11'] = 11
test_df['balance_avg'].loc[test_df['balance_avg']=='level 12'] = 12
test_df['balance_avg'].loc[test_df['balance_avg']=='level 13'] = 13
test_df['balance_avg'].loc[test_df['balance_avg']=='level 14'] = 14
test_df['balance_avg'].loc[test_df['balance_avg']=='level 15'] = 15
test_df['balance_avg'].loc[test_df['balance_avg']=='level 16'] = 16
test_df['balance_avg'].loc[test_df['balance_avg']=='level 17'] = 17
test_df['balance_avg'].loc[test_df['balance_avg']=='level 18'] = 18
test_df['balance_avg'].loc[test_df['balance_avg']=='level 19'] = 19
test_df['balance_avg'].loc[test_df['balance_avg']=='level 20'] = 20
test_df['balance_avg'].loc[test_df['balance_avg']=='level 21'] = 21

test_df['balance1'].loc[test_df['balance1']=='level 0'] = 0
test_df['balance1'].loc[test_df['balance1']=='level 1'] = 1
test_df['balance1'].loc[test_df['balance1']=='level 2'] = 2
test_df['balance1'].loc[test_df['balance1']=='level 3'] = 3
test_df['balance1'].loc[test_df['balance1']=='level 4'] = 4
test_df['balance1'].loc[test_df['balance1']=='level 5'] = 5
test_df['balance1'].loc[test_df['balance1']=='level 6'] = 6
test_df['balance1'].loc[test_df['balance1']=='level 7'] = 7
test_df['balance1'].loc[test_df['balance1']=='level 8'] = 8
test_df['balance1'].loc[test_df['balance1']=='level 9'] = 9
test_df['balance1'].loc[test_df['balance1']=='level 10'] = 10
test_df['balance1'].loc[test_df['balance1']=='level 11'] = 11
test_df['balance1'].loc[test_df['balance1']=='level 12'] = 12
test_df['balance1'].loc[test_df['balance1']=='level 13'] = 13

test_df['balance1'].loc[test_df['balance1']=='level 14'] = 14
test_df['balance1'].loc[test_df['balance1']=='level 15'] = 15
test_df['balance1'].loc[test_df['balance1']=='level 16'] = 16
test_df['balance1'].loc[test_df['balance1']=='level 17'] = 17
test_df['balance1'].loc[test_df['balance1']=='level 18'] = 18
test_df['balance1'].loc[test_df['balance1']=='level 19'] = 19
test_df['balance1'].loc[test_df['balance1']=='level 20'] = 20
test_df['balance1'].loc[test_df['balance1']=='level 21'] = 21

test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 0'] = 0
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 1'] = 1
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 2'] = 2
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 3'] = 3
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 4'] = 4
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 5'] = 5
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 6'] = 6
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 7'] = 7
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 8'] = 8
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 9'] = 9
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 10'] = 10
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 11'] = 11
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 12'] = 12
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 13'] = 13
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 14'] = 14
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 15'] = 15
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 16'] = 16
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 17'] = 17
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 18'] = 18
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 19'] = 19
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 20'] = 20
test_df['balance1_avg'].loc[test_df['balance1_avg']=='level 21'] = 21

test_df['balance2'].loc[test_df['balance2']=='level 0'] = 0
test_df['balance2'].loc[test_df['balance2']=='level 1'] = 1
test_df['balance2'].loc[test_df['balance2']=='level 2'] = 2
test_df['balance2'].loc[test_df['balance2']=='level 3'] = 3
test_df['balance2'].loc[test_df['balance2']=='level 4'] = 4
test_df['balance2'].loc[test_df['balance2']=='level 5'] = 5
test_df['balance2'].loc[test_df['balance2']=='level 6'] = 6
test_df['balance2'].loc[test_df['balance2']=='level 7'] = 7
test_df['balance2'].loc[test_df['balance2']=='level 8'] = 8
test_df['balance2'].loc[test_df['balance2']=='level 9'] = 9
test_df['balance2'].loc[test_df['balance2']=='level 10'] = 10
test_df['balance2'].loc[test_df['balance2']=='level 11'] = 11
test_df['balance2'].loc[test_df['balance2']=='level 12'] = 12
test_df['balance2'].loc[test_df['balance2']=='level 13'] = 13
test_df['balance2'].loc[test_df['balance2']=='level 14'] = 14
test_df['balance2'].loc[test_df['balance2']=='level 15'] = 15
test_df['balance2'].loc[test_df['balance2']=='level 16'] = 16
test_df['balance2'].loc[test_df['balance2']=='level 17'] = 17
test_df['balance2'].loc[test_df['balance2']=='level 18'] = 18
test_df['balance2'].loc[test_df['balance2']=='level 19'] = 19
test_df['balance2'].loc[test_df['balance2']=='level 20'] = 20
test_df['balance2'].loc[test_df['balance2']=='level 21'] = 21

test_df['balance2_avg'].loc[test_df['balance2_avg']=='level 0'] = 0
test_df['balance2_avg'].loc[test_df['balance2_avg']=='level 1'] = 1
test_df['balance2_avg'].loc[test_df['balance2_avg']=='level 2'] = 2
test_df['balance2_avg'].loc[test_df['balance2_avg']=='level 3'] = 3
test_df['balance2_avg'].loc[test_df['balance2_avg']=='level 4'] = 4
test_df['balance2_avg'].loc[test_df['balance2_avg']=='level 5'] = 5

test_df['service3'].loc[test_df['service3']=='category 1'] = 1
test_df['service3'].loc[test_df['service3']=='category 0'] = 0
test_df
test_df['product1_amount'].loc[test_df['product1_amount']=='level 1'] = 1
test_df['product1_amount'].loc[test_df['product1_amount']=='level 2'] = 2
test_df['product1_amount'].loc[test_df['product1_amount']=='level 3'] = 3
test_df['product1_amount'].loc[test_df['product1_amount']=='level 4'] = 4
test_df['product1_amount'].loc[test_df['product1_amount']=='level 5'] = 5
test_df['product1_amount'].loc[test_df['product1_amount']=='level 6'] = 6
test_df['product1_amount'].loc[test_df['product1_amount']=='level 7'] = 7

test_df['product2_amount'].loc[test_df['product2_amount']=='level 1'] = 1
test_df['product2_amount'].loc[test_df['product2_amount']=='level 2'] = 2
test_df['product2_amount'].loc[test_df['product2_amount']=='level 3'] = 3
test_df['product2_amount'].loc[test_df['product2_amount']=='level 4'] = 4
test_df['product2_amount'].loc[test_df['product2_amount']=='level 5'] = 5
test_df['product2_amount'].loc[test_df['product2_amount']=='level 6'] = 6
test_df['product2_amount'].loc[test_df['product2_amount']=='level 7'] = 7
test_df['product2_amount'].loc[test_df['product2_amount']=='level 8'] = 8
test_df['product2_amount'].loc[test_df['product2_amount']=='level 9'] = 9
test_df['product2_amount'].loc[test_df['product2_amount']=='level 10'] = 10
test_df['product2_amount'].loc[test_df['product2_amount']=='level 11'] = 11
test_df['product2_amount'].loc[test_df['product2_amount']=='level 12'] = 12
test_df['product2_amount'].loc[test_df['product2_amount']=='level 13'] = 13
test_df['product2_amount'].loc[test_df['product2_amount']=='level 14'] = 14
test_df['product2_amount'].loc[test_df['product2_amount']=='level 15'] = 15
test_df['product2_amount'].loc[test_df['product2_amount']=='level 16'] = 16
test_df['product2_amount'].loc[test_df['product2_amount']=='level 17'] = 17
test_df['product2_amount'].loc[test_df['product2_amount']=='level 18'] = 18
test_df['product2_amount'].loc[test_df['product2_amount']=='level 19'] = 19
test_df['product2_amount'].loc[test_df['product2_amount']=='level 20'] = 20
test_df['product2_amount'].loc[test_df['product2_amount']=='level 21'] = 21

test_df['product3_amount'].loc[test_df['product3_amount']=='level 1'] = 1
test_df['product3_amount'].loc[test_df['product3_amount']=='level 2'] = 2
test_df['product3_amount'].loc[test_df['product3_amount']=='level 3'] = 3

test_df['product4_amount'].loc[test_df['product4_amount']=='level 0'] = 0
test_df['product4_amount'].loc[test_df['product4_amount']=='level 1'] = 1

test_df['product5_amount'].loc[test_df['product5_amount']=='level 0'] = 0
test_df['product5_amount'].loc[test_df['product5_amount']=='level 1'] = 1

test_df['product6_amount'].loc[test_df['product6_amount']=='level 1'] = 1
test_df['product6_amount'].loc[test_df['product6_amount']=='level 2'] = 2
test_df['product6_amount'].loc[test_df['product6_amount']=='level 3'] = 3
test_df['product6_amount'].loc[test_df['product6_amount']=='level 4'] = 4
test_df['product6_amount'].loc[test_df['product6_amount']=='level 5'] = 5
test_df['product6_amount'].loc[test_df['product6_amount']=='level 6'] = 6
test_df['product6_amount'].loc[test_df['product6_amount']=='level 7'] = 7
test_df['product6_amount'].loc[test_df['product6_amount']=='level 8'] = 8
test_df['product6_amount'].loc[test_df['product6_amount']=='level 9'] = 9
test_df['product6_amount'].loc[test_df['product6_amount']=='level 10'] = 10
test_df['product6_amount'].loc[test_df['product6_amount']=='level 11'] = 11
test_df['product6_amount'].loc[test_df['product6_amount']=='level 12'] = 12
test_df['product6_amount'].loc[test_df['product6_amount']=='level 13'] = 13
test_df['product6_amount'].loc[test_df['product6_amount']=='level 14'] = 14
test_df['product6_amount'].loc[test_df['product6_amount']=='level 15'] = 15
test_df['product6_amount'].loc[test_df['product6_amount']=='level 16'] = 16
test_df['product6_amount'].loc[test_df['product6_amount']=='level 17'] = 17
test_df['product6_amount'].loc[test_df['product6_amount']=='level 18'] = 18
test_df['product6_amount'].loc[test_df['product6_amount']=='level 19'] = 19
test_df['product6_amount'].loc[test_df['product6_amount']=='level 20'] = 20
test_df['product6_amount'].loc[test_df['product6_amount']=='level 21'] = 21

删除类别较多的特征

train_df.drop(['province','provider','city','regist_type'],axis=1,inplace=True)
test_df.drop(['province','provider','city','regist_type'],axis=1,inplace=True)

更改数据类型

#改变数据类型
features = features = train_df[:1].drop('user',axis=1).columns
for i in features:
    if train_df[i].dtypes == object:
        train_df[i] = train_df[i].astype(int)
#改变数据类型
features = features = test_df[:1].drop('user',axis=1).columns
for i in features:
    if test_df[i].dtypes == object:
        test_df[i] = test_df[i].astype(int)

训练模型

#对模型进行训练并提交
drop_columns=["user","label"]
clf = lgb
train_count = train_df.shape[0]
#train_df = data_df[:train_count].copy().reset_index(drop=True)
#test_df = data_df[train_count:].copy().reset_index(drop=True)


features = train_df[:1].drop(drop_columns,axis=1).columns
train_x = train_df[features]
test_x = test_df[features]
test_x = test_x.fillna(method='ffill')
train_y = train_df['label']



train = np.zeros((train_x.shape[0], 1))
test = np.zeros((test_x.shape[0], 1))
    
nums = int(train_x.shape[0] * 0.90)
    
trn_x, trn_y, val_x, val_y = train_x[:nums], train_y[:nums], train_x[nums:], train_y[nums:]
    

train_matrix = clf.Dataset(trn_x, label=trn_y)
valid_matrix = clf.Dataset(val_x, label=val_y)
data_matrix  = clf.Dataset(train_x, label=train_y)
        
params = {
            'boosting_type': 'gbdt',
            'metric': {'binary_logloss', 'auc'},
            'min_child_weight': 5,
            'num_leaves': 2**6 ,#64
            'objective': 'binary',
            'feature_fraction': 0.9,
            'bagging_fraction': 0.7,
            'bagging_freq': 1,
            'learning_rate': 0.01,
            'seed': 520,
            'min_data_in_leaf': 500,
        }

model = clf.train(params, train_matrix, 50000, valid_sets=[train_matrix, valid_matrix], verbose_eval=500,early_stopping_rounds=1000)
model2 = clf.train(params, data_matrix, model.best_iteration)
val_pred = model.predict(val_x, num_iteration=model2.best_iteration).reshape(-1,1)
test_pred = model.predict(test_x, num_iteration=model2.best_iteration).reshape(-1,1)


lgb_train, lgb_test = val_pred，test_pred

提交文件

sub = pd.DataFrame()
sub['user'] = test_df.user
sub['prob'] = test_pred[:,0]
sub = sub.sort_values('user')
sub.to_csv('sub.csv',index=False)

本baseline只是简单的对数据进行了编码，还没有构造其他特征以及数据预处理等。

weixin_44015027
关注
  • 1
    点赞
  • 踩
  • 7
    收藏
    觉得还不错? 一键收藏
  • 10
    评论
翼支付杯大数据建模大赛-季军方案
weixin_45966291的博客
11-21 3596
翼支付杯大数据建模大赛-季军方案
决赛巅峰之战落幕，第二届翼支付杯大数据建模大赛完美收官
DataCastle
09-29 444
​历经2个月赛程，9月26日，“2020第二届翼支付杯大数据建模大赛”圆满落下帷幕，Cloud团队从10支决赛队伍中杀出重围，夺得本届大赛冠军！ 青春风暴，剑指未来 本届大赛由翼支付主办，三人行传媒集团股份有限公司承办，DataCastle数据城堡协办，并为参赛选手提供了16万现金奖励与翼支付复试直通卡/实习机会、证书，为校园新生力量提供才华施展的舞台和交流学习的通道，在实践中磨炼人工智能领域的专业能力，帮助学生完成从校园到社会的角色转变。大赛汇聚行业领域TOP媒体及运营平台，打造国内高校顶尖.
【数学建模】——【A题 信用风险识别问题】全面解析
最新发布
小李很执着的博客
08-02 8018
信用风险评价方法，包括数据预处理、特征选择、信用评分模型的构建及其比较、信用等级划分等多个方面，旨在提升信用风险评价的准确性和可靠性。
又一数据挖掘赛事，在校生专属，翼支付杯来了（直通实习机会）
Datawhale
07-22 536
Datawhale主办方：中国电信-翼支付，数据挖掘赛事为了积极研究探索“金融科技FinTech”技术并努力应用到实际业务中，挖掘更多金融科技在实际普惠金融业务的应用方案。由翼支付主...
16万大奖，名企offer，翼支付杯大数据建模大赛邀你来战！
DataCastle
08-06 494
第二届翼支付杯大数据建模大赛，自7月20日在DataCastle数据城堡开放报名通道以来，受到了社会各界的广泛关注！ 目前初赛阶段赛程过半，不少校园数据爱好者纷纷入场，新颖的赛题赛制和丰厚奖励让整个比赛的热度持续上升，高手云集，等你来战！ 关于“翼支付杯” 2019年，央行印发《金融科技（FinTech） 发展规划（2019—2021年）》，明确提出金融科技工作的指导思想、基本原则、发展目标、重点任务和保障措施。《规划》确定了六方面重点任务：一是加强金融科技战略部署；二是强...
Python构建信用卡客户风险识别模型
weixin_54707168的博客
08-17 1624
伴随着社会消费水平的提升,消费市场的繁荣促进了金融信贷行业的快速发展,以往由银行主导的信用卡业务不再是个体用户参与信贷活动的唯一选择,近几年由非金融银行机构推出的消费金融信贷方式受到了个人群体的亲睐。不幸的是，由于数据集保密问题，我们无法提供有关数据的原始功能和更多背景信息。基于收集到的数据，对每列进行缺失值统计，并没有存在缺失值，由于特征过多，存在很多与响应变量 无关的特征，因此需要进行特征提取。总体来看，随机森林的在测试集的正确率是0.73，其中正确预测了84个没有盗刷，18个发生盗刷的数据。
天池竞赛Baseline分享（2018广东工业智造大数据创新大赛-智能算法赛，线上0.921无调参）.zip
04-14
【标题】：“天池竞赛Baseline分享（2018广东工业智造大数据创新大赛-智能算法赛，线上0.921无调参）.zip”指的是阿里云天池平台举办的一次竞赛，主要关注的是2018年广东工业智能制造领域的大数据创新应用。...
“梧桐杯”中国移动大数据应用创新大赛 - 智慧金融赛道 Baseline.zip
04-14
“梧桐杯”中国移动大数据应用创新大赛 - 智慧金融赛道 Baseline.zip
山东省第二届数据应用创新创业大赛-主赛场-检验报告单识别-Baseline.zip
06-23
2、本项目适合计算机相关专业(如计科、人工智能、通信工程、自动化、电子信息等)的在校学生、老师或者企业员工下载学习，也适合小白学习进阶，当然也可作为毕设项目、课程设计、作业、项目初期立项演示等。...
第二届阿里巴巴大数据智能云上编程大赛-智联招聘人岗智能匹配-初赛数据-数据集
03-27
在第二届大赛中，智联招聘人岗智能匹配是初赛的重要课题，通过提供的数据集，参赛者需要构建模型以实现高效精准的人岗匹配。 本次大赛的数据集包含三个关键文件：table2_jd.csv（职位描述表）、table3_action.csv...
银行风险评估：数学建模实战
m0_52343631的博客
07-27 581
银行风险评估是金融领域中重要的应用之一。数学建模是银行风险评估的关键方法之一，它可以帮助我们处理大量的客户数据，提取有效特征，并进行准确的风险预测。通过Matlab编程，我们可以利用统计和机器学习工具箱来解决银行风险评估问题。本篇博客介绍了银行风险评估的数学建模方法，并提供了一个简单的实战案例来演示评估过程。希望本篇博客对您在银行风险评估方面有所启发。
风险大脑-支付风险识别天池大赛（三）快速建立模型（含调参思路）
little prince，blue coder
07-03 1987
         这一节开始建立模型，在建模之前，我们先梳理一下思路：   思路：   1、首先选出只带标签的数据（有风险和无风险的），进行分析，建立一个 recall rate 风险交易召回率较高的二分类模型。 2、接着在无标签的数据上应用此模型，打上标签，并与之前有标签的数据进行合并，形成最终完整训练样本。 3、将新的训练样本送入模型，得到最终模型。 4、应用测试数据，得到结果...
信用风险计量模型汇总
热门推荐
baidu_41605403的博客
10-18 1万+
信用风险计量模型的基本技术路线是，利用借款者的特征指标和宏观经济变量，收集这些特征指标和宏观变量的历史数据，并将其应用于预测违约借款人与履约借款人。预测模型旨在评估未知借款者将来是否还款的信用价值，将潜在借款者的特征值输入模型，从模型中输出信用价值评估，从而可对潜在借款人进行信用评估。 一般的评级方法可以分为专家经验判断法、参数模型和非参数模型。所谓的专家经验判断，就是相关专家根据主观经验...
用逻辑回归模型解决互联网金融信用风险问题
jacky_zhuyuanlu的博客
01-24 1983
（一）基础铺垫 逻辑回归（Logistic Regression） 针对因变量为分类变量而进行回归分析的一种统计方法，属于概率型非线性回归。 优点：算法易于实现和部署，执行效率和准确度高； 缺点：离散型的自变量数据需要通过生产虚拟变量的方式来使用； （二）底层原理及逻辑 在线性回归中，因变量是连续变量，那么线性回归能够根据因变量和自变量之间存在的线性关系来构造回归方程；但是，一旦
构建信用卡客户风险识别模型 + python 数据分析 + 图解
呵呵哒!的博客
12-25 1万+
构建信用卡客户风险识别模型 需求说明： 为了推进信用卡业务良性发展减少坏账风险，各大银行都进行了信用卡客户风险识别的相关工作，减少坏账风险，各大银行都进行了信用卡客户风险识别的相关工作。某银行研究的风险识别模型随时间推移不再适应业务发展需求，需要重新进行风险识别模型构建。 实验目标 （1） 掌握异常值的识别与处理方法。 （2） 构建信用卡用户风险分析关键特征。 （3） 掌握 K-Me...
银行数据仓库体系实践（18）--数据应用之信用风险建模
acumen_leo的博客
08-20 2366
风险应用作为银行的主要数据应用之一，也是数据仓库重点支持的数据应用方向，本节主要介绍了银行信用风险防控中的模型建设步骤以及应用，供各位参考。
【数学建模大作业】_“创新建模方法：银行收入优化的数学之道“_
记录我的学习历程
11-10 574
本研究通过两个模型对信用评分卡和阈值选择对银行收入的影响进行了深入分析。第一个模型采用模拟退火算法，强调全局搜索与简单实现的平衡。第二个模型基于暴力穷举算法，注重易理解性与实现的同时。然而，这两个模型均存在局限性，如规模适用性和多目标考虑。通过对两者优缺点的全面评估，本文提出了进一步改进与研究的方向，以构建更完备且适用于实际复杂问题的模型。
评论 10
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值