优惠券预测——数据探索1
import pandas as pd import numpy as np import seaborn as sns import matplotlib.pyplot as plt from datetime import date import datetime as dt from scipy import stats import warnings warnings.filterwarnings("ignore") %matplotlib inline
# 导入CSV文件
off_train = pd.read_csv('ccf_offline_stage1_train.csv', keep_default_na=False) off_train.columns = ['user_id', 'merchant_id', 'coupon_id', 'discount_rate', 'distance', 'date_received', 'date'] off_test = pd.read_csv('ccf_offline_stage1_test_revised.csv', keep_default_na=False) off_test.columns = ['user_id', 'merchant_id', 'coupon_id', 'discount_rate', 'distance', 'date_received'] on_train = pd.read_csv('ccf_online_stage1_train.csv', keep_default_na=False) on_train.columns = ['user_id', 'merchant_id', 'action', 'coupon_id', 'discount_rate', 'date_received', 'date']
off_train.head()
off_train.info()
off_test.head()
off_test.info()
# 领券日期范围 print('offline train date_received') print(off_train[off_train['date_received'] != 'null']['date_received'].min())# 非空日期 print(off_train[off_train['date_received'] != 'null']['date_received'].max()) print('online train date_received') print(on_train[on_train['date_received'] != 'null']['date_received'].min())# 非空日期 print(on_train[on_train['date_received'] != 'null']['date_received'].max()) print('offline test date_received') print(off_test[off_test['date_received'] != 'null']['date_received'].min())# 非空日期 print(off_test[off_test['date_received'] != 'null']['date_received'].max())
# 用券日期范围 print('offline train date') print(off_train[off_train['date'] != 'null']['date'].min())# 非空日期 print(off_train[off_train['date'] != 'null']['date'].max()) print('online train date') print(on_train[on_train['date'] != 'null']['date'].min())# 非空日期 print(on_train[on_train['date'] != 'null']['date'].max())
# 训练集与测试集id的重合度 # user_id off_train_user = off_train[['user_id']].copy().drop_duplicates() off_test_user = off_test[['user_id']].copy().drop_duplicates() on_train_user = on_train[['user_id']].copy().drop_duplicates() print('offline训练集用户ID数量') print(off_train_user.user_id.count()) print('online训练集用户ID数量') print(on_train_user.user_id.count()) print('offline测试集用户ID数量') print(off_test_user.user_id.count()) off_train_user['off_train_flag']=1 off_merge = off_test_user.merge(off_train_user, on='user_id', how="left").reset_index().fillna(0)# 索引,缺失值 print('offline训练集用户与测试集用户的重复数量') print(off_merge['off_train_flag'].sum()) print('offline训练集用户与测试集重复用户在总测试集用户中的占比') print(off_merge['off_train_flag'].sum()/off_test_user['user_id'].count()) on_train_user['on_train_flag']=1 on_merge = off_test_user.merge(on_train_user, on='user_id', how="left").reset_index().fillna(0) print('online训练集用户与测试集用户的重复数量') print(on_merge['on_train_flag'].sum()) print('online训练集用户与测试集重复用户在总测试集用户中的占比') print(on_merge['on_train_flag'].sum()/off_test_user['user_id'].count())
# plt.rcParams['figure.figsize'] = (25.0, 4.0) plt.title("Value Distribution", fontsize=24) plt.xlabel("Values", fontsize=14) plt.ylabel("Counts", fontsize=14) plt.tick_params(axis='both', labelsize=14) plt.xticks(size='small', rotation=68, fontsize=8) plt.plot(off_train['discount_rate'].value_counts(), linewidth=2) plt.show()