(学习赛)O2O优惠券使用预测

已于 2023-12-09 11:18:49 修改
阅读量945 收藏 16
点赞数
分类专栏： 论文-7算法50% 文章标签： 学习 数学建模
于 2021-07-27 07:20:10 首次发布
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/qq_45972323/article/details/119130423
版权 
#0.8055
import pandas as p
import numpy as n
import xgboost as xgb
def prepr(raw):
    pre=raw.copy()
    pre['num']=1
    #折扣率
    pre['MJ']=list(map(lambda x:1 if':'in str(x) else 0 ,pre['Discount_rate']))
    pre['JIAN']=list(map(lambda x:int(str(x).split(':')[1]) if":"in repr(x) else 0,pre['Discount_rate']))
    pre['MI_COST']=list(map(lambda x:int(str(x).split(':')[0]) if":"in repr(x) else 0,pre['Discount_rate']))
    pre['DISCOUNT']=list(map(lambda x:(float(str(x).split(':')[0])-float(str(x).split(':')[1]))/float(str(x).split(':')[0]) if":"in repr(x) else float(x),pre['Discount_rate']))
    #距离
    pre['Distance'].fillna(-1,inplace=True)
    pre['NUII_DISTANCE']=pre['Distance'].map(lambda x:1 if x == -1 else 0)
    #时间
    pre['DATE_RECEIVED']=p.to_datetime(pre['Date_received'],format='%Y%m%d')
    if'Date' in pre.columns.tolist():
        pre['DATE']=p.to_datetime(pre['Date'],format='%Y%m%d')
        pre['label'] = list(map(lambda y,x: 1 if (y-x).total_seconds()/(24*3600) <= 15 else 0,pre['DATE_RECEIVED'],pre['DATE']))
        pre['label'] = pre['label'].map(int)
    return pre

def construct_data(history,label):
    label_f=get_label_f(label)
    history_f=get_history_f(history,label)
    #构造数据集
    commom=list(set(label_f.columns.tolist())&set(history_f.columns.tolist()))
    data=p.concat([label_f, history_f.drop(commom, axis=1)],axis=1)
    #去重
    data.drop_duplicates(subset=None,keep='last',inplace=True)
    data.index=range(len(data))
    return data
def get_history_f(history,label):
    data=history.copy()
    data['Coupon_id']=data['Coupon_id'].map(int)
    data['Date_received']=data['Date_received'].map(int)
    h_f=label.copy()
    ###########################      用户
    keys=['User_id']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #用户领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x : str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    #用户  领满减数
    pivot=p.DataFrame(data[data['MJ']==1].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'MJ'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户领满减率
    h_f[prefixs + 'lu_MJ'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'MJ'],h_f[prefixs + 'received']))
    
    
    #用户15天内核销最大折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = max)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_max'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #用户15天内核销最小折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = min)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_min'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #用户15天内核销平均折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = n.mean)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_aver'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #用户15天内核销中位折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = n.median)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_median'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #用户15天内核销的最大距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = max)).rename(columns = {'Distance':prefixs + 'Distance_15_max'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #用户15天内核销的最小距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = min)).rename(columns = {'Distance':prefixs + 'Distance_15_min'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #用户15天内核销的平均距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = n.mean)).rename(columns = {'Distance':prefixs + 'Distance_15_mean'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #用户15天内核销的中位距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = n.median)).rename(columns = {'Distance':prefixs + 'Distance_15_median'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #用户15天内核销满减券减额最大值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = max)).rename(columns = {'JIAN': prefixs + "JIAN_max"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户15天内核销满减券减额最小值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = min)).rename(columns = {'JIAN': prefixs + "JIAN_min"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户15天内核销满减券减额平均值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.mean)).rename(columns = {'JIAN': prefixs + "JIAN_aver"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户15天内核销满减券减额中位值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.median)).rename(columns = {'JIAN': prefixs + "JIAN_median"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户15天内核销满减券最低消费最大值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = max)).rename(columns = {'MI_COST': prefixs + 'MI_COST_max'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户15天内核销满减券最低消费最小值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = min)).rename(columns = {'MI_COST': prefixs + 'MI_COST_min'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户15天内核销满减券最低消费平均值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = n.mean)).rename(columns = {'MI_COST': prefixs + 'MI_COST_aver'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户15天内核销满减券最低消费中位值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = n.median)).rename(columns = {'MI_COST': prefixs + 'MI_COST_medain'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    
    #################################         用户+商家
    keys=['User_id','Merchant_id']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #用户+商家领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+商家核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x : str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+商家核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    
    #用户+商家15天内核销最大折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = max)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_max'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #用户+商家15天内核销最小折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = min)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_min'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #用户+商家15天内核销平均折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = n.mean)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_aver'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #用户+商家15天内核销中位折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = n.median)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_median'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #用户+商家15天内核销满减券减额最大值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = max)).rename(columns = {'JIAN': prefixs + "JIAN_max"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户+商家15天内核销满减券减额最小值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = min)).rename(columns = {'JIAN': prefixs + "JIAN_min"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户+商家15天内核销满减券减额平均值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.mean)).rename(columns = {'JIAN': prefixs + "JIAN_aver"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户+商家15天内核销满减券减额中位值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.median)).rename(columns = {'JIAN': prefixs + "JIAN_median"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户+商家15天内核销满减券最低消费最大值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = max)).rename(columns = {'MI_COST': prefixs + 'MI_COST_max'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户+商家15天内核销满减券最低消费最小值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = min)).rename(columns = {'MI_COST': prefixs + 'MI_COST_min'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户+商家15天内核销满减券最低消费平均值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = n.mean)).rename(columns = {'MI_COST': prefixs + 'MI_COST_aver'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #用户+商家15天内核销满减券最低消费中位值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = n.median)).rename(columns = {'MI_COST': prefixs + 'MI_COST_medain'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    
    #################################           用户+优惠券
    keys=['User_id','Coupon_id']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #用户+优惠券领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+优惠券核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x : str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+优惠券核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    #################################             用户+折扣率
    keys=['User_id','DISCOUNT']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #用户+折扣率 领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+折扣率 核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x:str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+折扣率 核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    
    #################################            用户+日期
    keys=['User_id','DATE_RECEIVED']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #用户+日期领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+日期核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x : str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+日期核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    #################################                用户+商家+优惠券
    keys=['User_id','Merchant_id','Coupon_id']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #用户+商家+优惠券 领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+商家+优惠券 核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x : str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+商家+优惠券 核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    #################################                   用户+商家+日期
    keys=['User_id','Merchant_id','DATE_RECEIVED']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #  用户+商家+日期领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    # 用户+商家+日期 核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x : str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    # 用户+商家+日期 核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    #################################                   用户+优惠券+日期
    keys=['User_id','Coupon_id','DATE_RECEIVED']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #用户+优惠券+日期 领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+优惠券+日期 核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x : str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #用户+优惠券+日期 核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    
    #################################        商家
    keys=['Merchant_id']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #商家领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #商家核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x : str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #商家核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    #商家15天内核销的最大距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = max)).rename(columns = {'Distance':prefixs + 'Distance_15_max'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #商家15天内核销的最小距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = min)).rename(columns = {'Distance':prefixs + 'Distance_15_min'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #商家15天内核销的平均距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = n.mean)).rename(columns = {'Distance':prefixs + 'Distance_15_aver'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #商家15天内核销的中位距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = n.median)).rename(columns = {'Distance':prefixs + 'Distance_15_median'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #商家15天内核销最大折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = max)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_max'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #商家15天内核销最小折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = min)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_min'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #商家15天内核销平均折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = n.mean)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_aver'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #商家15天内核销中位折扣率
    pivot = p.DataFrame(data[data['label']==1].pivot_table( index = keys, values ='DISCOUNT',aggfunc = n.median)).rename(columns = {'DISCOUNT':prefixs + 'DISCOUNT_15_median'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how ='left')
    #商家15天内核销满减券减额最大值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = max)).rename(columns = {'JIAN': prefixs + "JIAN_max"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #商家15天内核销满减券减额最小值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = min)).rename(columns = {'JIAN': prefixs + "JIAN_min"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #商家15天内核销满减券减额平均值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.mean)).rename(columns = {'JIAN': prefixs + "JIAN_aver"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #商家15天内核销满减券减额中位值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.median)).rename(columns = {'JIAN': prefixs + "JIAN_median"}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #商家15天内核销满减券最低消费最大值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = max)).rename(columns = {'MI_COST': prefixs + 'MI_COST_max'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #商家15天内核销满减券最低消费最小值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = min)).rename(columns = {'MI_COST': prefixs + 'MI_COST_min'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #商家15天内核销满减券最低消费平均值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = n.mean)).rename(columns = {'MI_COST': prefixs + 'MI_COST_aver'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    #商家15天内核销满减券最低消费中位值
    pivot = p.DataFrame(data[data['label']==1].pivot_table(index = keys, values = 'MI_COST', aggfunc = n.median)).rename(columns = {'MI_COST': prefixs + 'MI_COST_medain'}).reset_index()
    h_f = p.merge(h_f, pivot, on = keys, how = 'left')
    
    #################################             商家+优惠券
    keys=['Merchant_id','Coupon_id']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #商家+优惠券领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #商家+优惠券核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x:str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #商家+优惠券核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    #################################             商家+折扣率
    keys=['Merchant_id','DISCOUNT']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #商家+折扣率领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #商家+折扣率核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x:str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #商家+折扣率核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    
    #################################             商家+日期
    keys=['Merchant_id','DATE_RECEIVED']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #商家+日期领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #商家+日期核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x:str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #商家+日期核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    #################################             商家+优惠券+日期
    keys=['Merchant_id','Coupon_id','DATE_RECEIVED']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #商家+优惠券+日期 领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #商家+优惠券+日期核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x:str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #商家+优惠券+日期核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    
    #################################               优惠券
    keys=['Coupon_id']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #优惠券领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #优惠券核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x : str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #优惠券核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    
    #优惠券15天内核销的最大距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = max)).rename(columns = {'Distance':prefixs + 'Distance_15_max'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #优惠券15天内核销的最小距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = min)).rename(columns = {'Distance':prefixs + 'Distance_15_min'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #优惠券15天内核销的平均距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = n.mean)).rename(columns = {'Distance':prefixs + 'Distance_15_mean'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #优惠券15天内核销的中位距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = n.median)).rename(columns = {'Distance':prefixs + 'Distance_15_median'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #################################               优惠券+日期
    keys=['Coupon_id','DATE_RECEIVED']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #优惠券+日期领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #优惠券+日期核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x : str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #优惠券+日期核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    
    #################################             折扣率
    keys=['DISCOUNT']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #折扣率领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #折扣率核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x:str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #折扣率核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    
    #################################日期
    keys=['DATE_RECEIVED']
    prefixs='history_field_'+'_'.join(keys)+'_'
    #当日领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #当日核销数
    pivot=p.DataFrame(data[data['Date'].map(lambda x : str(x) !='nan')].pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received_use'}).reset_index()
    h_f=p.merge(h_f,pivot,on=keys,how='left')
    #当日核销率
    h_f[prefixs + 'lu_use'] = list(map(lambda x,y: x/y if y!=0 else 0 ,h_f[prefixs + 'received_use'],h_f[prefixs + 'received']))
    
    #当日15天内核销的最大距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = max)).rename(columns = {'Distance':prefixs + 'Distance_15_max'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #当日15天内核销的最小距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = min)).rename(columns = {'Distance':prefixs + 'Distance_15_min'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #当日15天内核销的平均距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = n.mean)).rename(columns = {'Distance':prefixs + 'Distance_15_mean'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #当日15天内核销的中位距离
    pivot =p.DataFrame(data[data['label']==1].pivot_table( index = keys, values = 'Distance',aggfunc = n.median)).rename(columns = {'Distance':prefixs + 'Distance_15_median'}).reset_index()
    h_f = p.merge(h_f, pivot, how ='left', on = keys)
    #用户距离正反排序
    h_f['label_User_distance_true_rank']=h_f.groupby('User_id')['Distance'].rank(ascending=True)
    h_f['label_User_distance_False_rank']=h_f.groupby('User_id')['Distance'].rank(ascending=False)  
    
    #用户折扣正反排序
    h_f['label_User_discount_rate_true_rank']=h_f.groupby('User_id')['DISCOUNT'].rank(ascending=True)
    h_f['label_User_discount_rate_False_rank']=h_f.groupby('User_id')['DISCOUNT'].rank(ascending=False)
  
    #用户领券日期正反排序
    h_f['label_User_date_received_true_rank']=h_f.groupby('User_id')['DATE_RECEIVED'].rank(ascending=True)
    h_f['label_User_date_received_False_rank']=h_f.groupby('User_id')['DATE_RECEIVED'].rank(ascending=False)
    
    
    
    ####
    #商家距离正反排序
    h_f['label_Merchant_distance_true_rank']=h_f.groupby('Merchant_id')['Distance'].rank(ascending=True)
    h_f['label_Merchant_distance_False_rank']=h_f.groupby('Merchant_id')['Distance'].rank(ascending=False)  
    
    #商家折扣正反排序
    h_f['label_Merchant_discount_rate_true_rank']=h_f.groupby('Merchant_id')['DISCOUNT'].rank(ascending=True)
    h_f['label_Merchant_discount_rate_False_rank']=h_f.groupby('Merchant_id')['DISCOUNT'].rank(ascending=False)
  
    #商家领券日期正反排序
    h_f['label_Merchant_date_received_true_rank']=h_f.groupby('Merchant_id')['DATE_RECEIVED'].rank(ascending=True)
    h_f['label_Merchant_date_received_False_rank']=h_f.groupby('Merchant_id')['DATE_RECEIVED'].rank(ascending=False)
    
    #####
    
    ############################################~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    #优惠券距离正反排序
    h_f['label_Coupon_distance_true_rank']=h_f.groupby('Coupon_id')['Distance'].rank(ascending=True)
    h_f['label_Coupon_distance_False_rank']=h_f.groupby('Coupon_id')['Distance'].rank(ascending=False)  
    
    #优惠券折扣正反排序
    h_f['label_Coupon_discount_rate_true_rank']=h_f.groupby('Coupon_id')['DISCOUNT'].rank(ascending=True)
    h_f['label_Coupon_discount_rate_False_rank']=h_f.groupby('Coupon_id')['DISCOUNT'].rank(ascending=False)
  
    #优惠券领券日期正反排序
    h_f['label_Coupon_date_received_true_rank']=h_f.groupby('Coupon_id')['DATE_RECEIVED'].rank(ascending=True)
    h_f['label_Coupon_date_received_False_rank']=h_f.groupby('Coupon_id')['DATE_RECEIVED'].rank(ascending=False)
    
    
    
    h_f.fillna(0,downcast='infer',inplace=True)
    return h_f  
def get_label_f(label):
    data=label.copy()
    data['Coupon_id']=data['Coupon_id'].map(int)
    data['Date_received']=data['Date_received'].map(int)
    l_f= label.copy()
    ###################################用户
    keys=['User_id']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #每个用户领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    
    
    #用户领券的最大距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = max)).rename(columns = {'Distance': prefixs + "Distance_max"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户领券的最小距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = min)).rename(columns = {'Distance': prefixs + "Distance_min"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户领券的最小距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = n.mean)).rename(columns = {'Distance': prefixs + "Distance_aver"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户领券的最小距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = n.median)).rename(columns = {'Distance': prefixs + "Distance_median"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户优惠券折扣率最大值
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = max)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_max'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户优惠券折扣率最小值
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = min)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_min'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户优惠券折扣率平均数
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = n.mean)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_mean'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户优惠券折扣率中位数
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = n.median)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_median'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户领满减券最低消费最大值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=max)).rename(columns = {'MI_COST': prefixs + 'MI_COST_max'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户领满减券最低消费最小值 
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=min)).rename(columns = {'MI_COST': prefixs + 'MI_COST_min'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户领满减券最低消费平均数 
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=n.mean)).rename(columns = {'MI_COST': prefixs + 'MI_COST_aver'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户领满减券最低消费中位数
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=n.median)).rename(columns = {'MI_COST': prefixs + 'MI_COST_median'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户领满减券减额最大值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = max)).rename(columns = {'JIAN': prefixs + "JIAN_max"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户领满减券减额最小值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = min)).rename(columns = {'JIAN': prefixs + "JIAN_min"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户领满减券减额平均值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.mean)).rename(columns = {'JIAN': prefixs + "JIAN_aver"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户领满减券减额中位数
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.median)).rename(columns = {'JIAN': prefixs + "JIAN_median"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    
    tmp = data[keys+['DATE_RECEIVED']].sort_values(['DATE_RECEIVED'],ascending=True)
    #用户第一次领券
    first = tmp.drop_duplicates(keys,keep="first")
    first[prefixs+"is_first_received"]=1
    l_f = p.merge(l_f,first,on=keys+['DATE_RECEIVED'],how="left")
    #用户最后一次领券
    last = tmp.drop_duplicates(keys,keep="last")
    last[prefixs+"is_last_received"] = 1
    l_f = p.merge(l_f,last,on=keys+['DATE_RECEIVED'],how="left")
    #################################用户+商家
    keys=['User_id','Merchant_id']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #用户+商家领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    #用户+商家优惠券折扣率最大值
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = max)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_max'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户+商家优惠券折扣率最小值
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = min)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_min'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户+商家优惠券折扣率平均数
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = n.mean)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_mean'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户+商家优惠券折扣率中位数
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = n.median)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_median'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户+商家领满减券最低消费最大值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=max)).rename(columns = {'MI_COST': prefixs + 'MI_COST_max'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户+商家领满减券最低消费最小值 
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=min)).rename(columns = {'MI_COST': prefixs + 'MI_COST_min'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户+商家领满减券最低消费平均数 
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=n.mean)).rename(columns = {'MI_COST': prefixs + 'MI_COST_aver'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户+商家领满减券最低消费中位数
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=n.median)).rename(columns = {'MI_COST': prefixs + 'MI_COST_median'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户+商家领满减券减额最大值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = max)).rename(columns = {'JIAN': prefixs + "JIAN_max"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户+商家领满减券减额最小值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = min)).rename(columns = {'JIAN': prefixs + "JIAN_min"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户+商家领满减券减额平均值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.mean)).rename(columns = {'JIAN': prefixs + "JIAN_aver"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #用户+商家领满减券减额中位数
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.median)).rename(columns = {'JIAN': prefixs + "JIAN_median"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    tmp = data[keys+['DATE_RECEIVED']].sort_values(['DATE_RECEIVED'],ascending=True)
    #用户+商家第一次领券
    first = tmp.drop_duplicates(keys,keep="first")
    first[prefixs+"is_first_received"]=1
    l_f = p.merge(l_f,first,on=keys+['DATE_RECEIVED'],how="left")
    #用户+商家最后一次领券
    last = tmp.drop_duplicates(keys,keep="last")
    last[prefixs+"is_last_received"] = 1
    l_f = p.merge(l_f,last,on=keys+['DATE_RECEIVED'],how="left")
    #################################用户+优惠券
    keys=['User_id','Coupon_id']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #用户+优惠券领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    #################################用户+折扣率
    keys=['User_id','DISCOUNT']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #用户+折扣率领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    
    tmp = data[keys+['DATE_RECEIVED']].sort_values(['DATE_RECEIVED'],ascending=True)
    #用户+折扣率第一次领券
    first = tmp.drop_duplicates(keys,keep="first")
    first[prefixs+"is_first_received"]=1
    l_f = p.merge(l_f,first,on=keys+['DATE_RECEIVED'],how="left")
    #用户+折扣率最后一次领券
    last = tmp.drop_duplicates(keys,keep="last")
    last[prefixs+"is_last_received"] = 1
    l_f = p.merge(l_f,last,on=keys+['DATE_RECEIVED'],how="left")
    #################################用户+日期
    keys=['User_id','DATE_RECEIVED']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #用户+日期领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    #################################用户+商家+优惠券
    keys=['User_id','Merchant_id','Coupon_id']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #用户+商家+优惠券领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    #################################用户+商家+日期
    keys=['User_id','Merchant_id','DATE_RECEIVED']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #用户+商家+日期领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    #################################用户+优惠券+日期
    keys=['User_id','Coupon_id','DATE_RECEIVED']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #用户+优惠券+日期领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    #################################商家
    keys=['Merchant_id']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #商家被领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    #商家被领券的最大距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = max)).rename(columns = {'Distance': prefixs + "Distance_max"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家被领券的最小距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = min)).rename(columns = {'Distance': prefixs + "Distance_min"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家被领券的最小距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = n.mean)).rename(columns = {'Distance': prefixs + "Distance_aver"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家被领券的最小距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = n.median)).rename(columns = {'Distance': prefixs + "Distance_median"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家优惠券折扣率最大值
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = max)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_max'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家优惠券折扣率最小值
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = min)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_min'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家优惠券折扣率平均数
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = n.mean)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_mean'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家优惠券折扣率中位数
    pivot = p.DataFrame(data.pivot_table( index = keys, values = 'DISCOUNT', aggfunc = n.median)).rename(columns = {'DISCOUNT': prefixs + 'DISCOUNT_median'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家被领满减券最低消费最大值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=max)).rename(columns = {'MI_COST': prefixs + 'MI_COST_max'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家被领满减券最低消费最小值 
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=min)).rename(columns = {'MI_COST': prefixs + 'MI_COST_min'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家被领满减券最低消费平均数 
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=n.mean)).rename(columns = {'MI_COST': prefixs + 'MI_COST_aver'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家被领满减券最低消费中位数
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'MI_COST',aggfunc=n.median)).rename(columns = {'MI_COST': prefixs + 'MI_COST_median'}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家被领满减券减额最大值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = max)).rename(columns = {'JIAN': prefixs + "JIAN_max"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家被领满减券减额最小值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = min)).rename(columns = {'JIAN': prefixs + "JIAN_min"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家被领满减券减额平均值
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.mean)).rename(columns = {'JIAN': prefixs + "JIAN_aver"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #商家被领满减券减额中位数
    pivot = p.DataFrame(data[data['MJ']==1].pivot_table(index = keys, values = 'JIAN', aggfunc = n.median)).rename(columns = {'JIAN': prefixs + "JIAN_median"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    tmp = data[keys+['DATE_RECEIVED']].sort_values(['DATE_RECEIVED'],ascending=True)
    #商家被第一次领券
    first = tmp.drop_duplicates(keys,keep="first")
    first[prefixs+"is_first_received"]=1
    l_f = p.merge(l_f,first,on=keys+['DATE_RECEIVED'],how="left")
    #商家被最后一次领券
    last = tmp.drop_duplicates(keys,keep="last")
    last[prefixs+"is_last_received"] = 1
    l_f = p.merge(l_f,last,on=keys+['DATE_RECEIVED'],how="left")
    #################################商家+优惠券
    keys=['Merchant_id','Coupon_id']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #商家+优惠券领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    
    #################################商家+折扣率
    keys=['Merchant_id','DISCOUNT']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #商家+折扣率领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    
    tmp = data[keys+['DATE_RECEIVED']].sort_values(['DATE_RECEIVED'],ascending=True)
    #商家+折扣率第一次领券
    first = tmp.drop_duplicates(keys,keep="first")
    first[prefixs+"is_first_received"]=1
    l_f = p.merge(l_f,first,on=keys+['DATE_RECEIVED'],how="left")
    #商家+折扣率最后一次领券
    last = tmp.drop_duplicates(keys,keep="last")
    last[prefixs+"is_last_received"] = 1
    l_f = p.merge(l_f,last,on=keys+['DATE_RECEIVED'],how="left")
    #################################商家+日期
    keys=['Merchant_id','DATE_RECEIVED']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #商家+日期领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    #################################商家+优惠券+日期
    keys=['Merchant_id','Coupon_id','DATE_RECEIVED']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #商家+优惠券+日期领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    #################################优惠券
    keys=['Coupon_id']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #优惠券领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    
    #优惠券被领券的最大距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = max)).rename(columns = {'Distance': prefixs + "Distance_max"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #优惠券被领券的最小距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = min)).rename(columns = {'Distance': prefixs + "Distance_min"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #优惠券被领券的最小距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = n.mean)).rename(columns = {'Distance': prefixs + "Distance_aver"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #优惠券被领券的最小距离
    pivot = p.DataFrame(data.pivot_table(index = keys, values = 'Distance', aggfunc = n.median)).rename(columns = {'Distance': prefixs + "Distance_median"}).reset_index()
    l_f = p.merge(l_f, pivot, on = keys, how = 'left')
    #################################优惠券+日期
    keys=['Coupon_id','DATE_RECEIVED']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #优惠券+日期领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    #################################折扣率
    keys=['DISCOUNT']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #折扣率 被领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    tmp = data[keys+['DATE_RECEIVED']].sort_values(['DATE_RECEIVED'],ascending=True)
    #折扣率 被第一次领券
    first = tmp.drop_duplicates(keys,keep="first")
    first[prefixs+"is_first_received"]=1
    l_f = p.merge(l_f,first,on=keys+['DATE_RECEIVED'],how="left")
    #折扣率 被最后一次领券
    last = tmp.drop_duplicates(keys,keep="last")
    last[prefixs+"is_last_received"] = 1
    l_f = p.merge(l_f,last,on=keys+['DATE_RECEIVED'],how="left")
    #################################日期
    keys=['DATE_RECEIVED']
    prefixs='label_field_'+'_'.join(keys)+'_'
    #当日领券数
    pivot=p.DataFrame(data.pivot_table(index=keys,values='num',aggfunc=len)).rename(columns={'num':prefixs+'received'}).reset_index()
    l_f=p.merge(l_f,pivot,on=keys,how='left')
    
    
    #用户距离正反排序
    l_f['label_User_distance_true_rank']=l_f.groupby('User_id')['Distance'].rank(ascending=True)
    l_f['label_User_distance_False_rank']=l_f.groupby('User_id')['Distance'].rank(ascending=False)  
    
    #用户折扣正反排序
    l_f['label_User_discount_rate_true_rank']=l_f.groupby('User_id')['DISCOUNT'].rank(ascending=True)
    l_f['label_User_discount_rate_False_rank']=l_f.groupby('User_id')['DISCOUNT'].rank(ascending=False)
  
    
    
    
    ####
    #商家距离正反排序
    l_f['label_Merchant_distance_true_rank']=l_f.groupby('Merchant_id')['Distance'].rank(ascending=True)
    l_f['label_Merchant_distance_False_rank']=l_f.groupby('Merchant_id')['Distance'].rank(ascending=False)  
    
    #商家折扣正反排序
    l_f['label_Merchant_discount_rate_true_rank']=l_f.groupby('Merchant_id')['DISCOUNT'].rank(ascending=True)
    l_f['label_Merchant_discount_rate_False_rank']=l_f.groupby('Merchant_id')['DISCOUNT'].rank(ascending=False)

    #####
    
    ############################################~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    #优惠券距离正反排序
    l_f['label_Coupon_distance_true_rank']=l_f.groupby('Coupon_id')['Distance'].rank(ascending=True)
    l_f['label_Coupon_distance_False_rank']=l_f.groupby('Coupon_id')['Distance'].rank(ascending=False)  
    
    #优惠券折扣正反排序
    l_f['label_Coupon_discount_rate_true_rank']=l_f.groupby('Coupon_id')['DISCOUNT'].rank(ascending=True)
    l_f['label_Coupon_discount_rate_False_rank']=l_f.groupby('Coupon_id')['DISCOUNT'].rank(ascending=False)
  
    #优惠券领券日期正反排序
    l_f['label_Coupon_date_received_true_rank']=l_f.groupby('Coupon_id')['DATE_RECEIVED'].rank(ascending=True)
    l_f['label_Coupon_date_received_False_rank']=l_f.groupby('Coupon_id')['DATE_RECEIVED'].rank(ascending=False)
    
    
    l_f.fillna(0,downcast='infer',inplace=True)
    return l_f

def model_xgb(train,test):
    params={
              "booster": 'gbtree',
              'objective': 'binary:logistic',
              'eval_metric': 'auc',
              'silent': 0,#(静默模式,1开0关)
              'eta': 0.01,#(0.01~0.2,,,0.01)
              'max_depth': 5,#(3~10,,,6)
              'min_child_weight': 1,
              'gamma': 0,
              'lambda': 1,
              'colsample_bylevel': 0.7,#(作用与subsample相似)
              'colsample_bytree': 0.7,#(0.5~1)
              'subsample': 0.9,#(0.5~1)
              'scale_pos_weight': 1,#(算法更快收敛)
        }
    #数据集
    dtrain=xgb.DMatrix(train.drop(['User_id','Coupon_id','Merchant_id', 'Discount_rate', 'Date', 'DATE_RECEIVED','Date_received','label','DATE'],axis=1),label=train['label'])
    dtest=xgb.DMatrix(test.drop(['User_id','Coupon_id','Merchant_id', 'Discount_rate',  'DATE_RECEIVED','Date_received'],axis=1))
    #训练
    watchlist=[(dtrain,'train')]
    model=xgb.train(params,dtrain,3,watchlist)
    #1300轮
    #预测
    predict=model.predict(dtest)
    #结果
    predict=p.DataFrame(predict,columns=['prob'])
    result=p.concat([test[['User_id','Coupon_id','Date_received']],predict],axis=1)
    #特征的重要性
    feat_importance = p.DataFrame(columns=['feature_name', 'importance'])
    feat_importance['feature_name'] = model.get_score().keys()
    feat_importance['importance'] = model.get_score().values()
    feat_importance.sort_values(['importance'], ascending=False, inplace=True)
    return result,feat_importance 

if __name__ =='__main__':
    #原数据
    raw_train=p.read_csv("D:/1A动手实践/Python/1O2O/ccf_offline_stage1_train.csv")
    raw_test=p.read_csv("D:/1A动手实践/Python/1O2O/ccf_offline_stage1_test_revised.csv")
    #预处理
    prepr_train=prepr(raw_train)
    prepr_test=prepr(raw_test)
    #划分区间
    #训练集 历史，中间，标签区间
    train_history=prepr_train[prepr_train['DATE_RECEIVED'].isin(p.date_range('2016/3/2',periods=60))]
    train_label=prepr_train[prepr_train['DATE_RECEIVED'].isin(p.date_range('2016/5/16',periods=31))]
    #验证集 历史，中间，标签区间
    verification_history=prepr_train[prepr_train['DATE_RECEIVED'].isin(p.date_range('2016/1/16',periods=60))]
    verification_label=prepr_train[prepr_train['DATE_RECEIVED'].isin(p.date_range('2016/3/31',periods=31))]
    #测试集 历史，中间，标签区间
    test_history=prepr_train[prepr_train['DATE_RECEIVED'].isin(p.date_range('2016/4/17',periods=60))]
    test_label=prepr_test.copy()
    #构造数据集
    complete_train=construct_data(train_history,train_label)
    complete_verification=construct_data(verification_history,verification_label)
    complete_test=construct_data(test_history,test_label)
    good_train=p.concat([complete_train,complete_verification],axis=0)
    
    
    result,feat_importance=model_xgb(good_train,complete_test)
    result.to_csv("10-17-2.csv",index=False,header=None)
大数据驱动AI
关注
  • 0
    点赞
  • 踩
  • 16
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
专栏目录
O2O优惠券使用预测项目总结
Hikikaya的博客
09-27 5263
O2O优惠券使用预测笔记前言项目介绍数据评价方式赛题分析特征选取模型选取过程及代码总结合理的创建标题，有助于目录的生成如何改变文本的样式插入链接与图片如何插入一段漂亮的代码片生成一个适合你的列表创建一个表格设定内容居中、居左、居右SmartyPants创建一个自定义列表如何创建一个注脚注释也是必不可少的KaTeX数学公式新的甘特图功能，丰富你的文章UML 图表FLowchart流程图导出与导入导出导入 前言 笔者希望通过本篇文章总结阿里天池O2O优惠券使用预测比赛的实践成果，以期不断提高自身数据分析和数据挖
天池比赛：o2o优惠券使用预测
u013402321的博客
01-22 3543
一、比赛背景O2O：全称Online To Offline，线上线下电子商务，是把线上的消费者带到现实的商店中去：在线支付线下商品、服务，再到线下去享受服务。通过打折（例如团购）、提供信息、服务（例如预定）等方式，把线下商店的消息推送给互联网用户，从而将他们转换为自己的线下客户。这样线下服务就可以用线上来揽客，消费者可以用线上来筛选服务，还有成交可以在线结算。通俗点说，你在网上买个优惠券去麦当劳消费
【机器学习】XGB实战—O2O优惠券使用预测
littlemichelle
12-28 598
比赛地址：https://tianchi.aliyun.com/competition/entrance/231593/introduction。论坛：https://tianchi.aliyun.com/competition/entrance/231593/forum。数据字段需特别说明的，received_data 点击数据；used_data 消费数据；
阿里云天池O2O优惠券预测之模型交叉验证部分Python代码
qq_44386182的博客
09-29 537
算法模型比较# 修改pandas默认的现实设置"""按照： 评价指标 验证方式 学习曲线 结果分析，模型选择 模型调参 的步骤对模型进行选择，调优"""#########部分SKLearn 集成的算法########################SKLearn 集成的算法###########################全局参数#############################################目录定义#################################
Sklearn：天池新人实战赛o2o优惠券使用预测 part1
あずにゃん梓喵的博客
07-23 1339
日萌社 人工智能AI：Keras PyTorch MXNet TensorFlow PaddlePaddle 深度学习实战（不定时更新） 阿里云官网：天池新人实战赛o2o优惠券使用预测 数据集下载链接：https://pan.baidu.com/s/13OtaUv6j4x8dD7cgD4sL5g提取码：7tze 数据 本赛题提供用户在2016年1月1日至2016年6月30日之...
O2O优惠券使用预测 阿里天池新人实战赛 总结
最新发布
qq_51866108的博客
10-08 1060
首先对提供的数据进行简单的预处理，接着时间窗口划分法将数据集进行划分和打标，最后进行特征工程处理，即特征提取和模型建立与参数优化，进而完成整个数据的预测。目标：预测投放的优惠券是否核销（使用）实质：分类问题，也就是：是/否消费优惠券首先需要说一说GBDT，它是一种基于boosting增强策略的加法模型，训练的时候采用前向分布算法进行贪婪的学习，每次迭代都学习一棵CART树来拟合之前 t-1 棵树的预测结果与训练样本真实值的残差。
【深度之眼】三：分类 —— 算法实战 —— o2o 优惠券使用预测比赛【天池】
InitialHeart2021的博客
01-01 384
o2o 优惠券使用预测比赛_初级
Xgboost实践 | 第一名天池o2o优惠券的使用预测思路完整版
AI蜗牛车
05-31 3587
前言本篇我们来分享一个天池上面的机器学习比赛:o2o优惠券的使用预测（目前第一名auc:0.81，本篇:0.786, 待优化），看完本篇希望大家掌握xgboost的用法以及提取特征的思路。这篇文章是笔者很早以前写的，看到很多小伙伴在讨论交流，说明大家还是学习到不少东西，故再分享一次，希望对大家有帮助，有任何问题欢迎评论区交流，或去如下的笔者csdn博客交流，其上已有很多小...
天池o2o优惠券使用预测（入门）
JhonyLin的博客
04-23 3441
一、前言 近期学习了一下天池中o2o优惠券使用预测的学习赛，主要任务是通过分析建模，精准预测用户是否会在规定时间内使用相应优惠券。这次的参与主要是学习为主，牛刀小试。 二、解决方案 数据分析：对于给定的数据集进行分析处理。 特征工程：挖掘出更具代表性的特征。 模型建立：使用随机梯度下降法进行建模（SGDClassifier） 三、功能实现 1、导入相关的库 #导入常规用的库 import n...
天池新人实战赛o2o优惠券使用预测-数据集
04-01
"天池新人实战赛o2o优惠券使用预测-数据集"是一个全面了解和实践数据分析流程的良好平台，涵盖了数据获取、数据清洗、特征工程、模型训练、预测及结果提交等多个环节，对于提升IT新人在大数据分析和机器学习领域的...
天池新人实战赛o2o优惠券使用预测-lingyunmu-数据集
04-01
该数据集来源于阿里云天池平台举办的一场新人实战赛，主题是“o2o优惠券使用预测”。o2o（Online To Offline）是指线上到线下的商业模式，常见于电商平台和生活服务类应用中，通过发放优惠券刺激消费者在线下消费。...
天池o2o优惠券使用预测-数据集
03-29
【标题】"天池o2o优惠券使用预测-...总的来说，"天池o2o优惠券使用预测-数据集"是一个涉及数据挖掘、机器学习和业务理解的综合挑战，参赛者需要结合业务背景，运用多种技术手段构建能够准确预测优惠券使用情况的模型。
将python项目部署在一台服务器上
qq_45972323的博客
10-06 2296
我的是物理服务器：这是将服务器硬件直接放置在您自己的数据中心或机房的传统方法。这种方法需要更多的设备和维护工作，但提供更高的灵活性和控制权。使用Docker，可以将Python程序及其依赖项打包成一个镜像，并在服务器上运行该镜像。优点：Docker部署Python程序，可以提供更好的可移植性和可扩展性、可以在不同的服务器和环境中轻松地运行相同的镜像。优点：服务器less技术部署Python程序，可以提供更高的可伸缩。容器化技术是将应用程序及其依赖项以容器的形式打包和部署的一种方法。
简单算法部署
qq_45972323的博客
05-25 1646
算法部署
C++入门_9通讯录管理系统
qq_45972323的博客
01-30 965
C++笔记_9通讯录管理系统系统需求创建项目菜单功能添加联系人显示联系人删除联系人查找联系人修改联系人清空联系人退出功能 系统需求 创建项目 菜单功能 添加联系人 显示联系人 删除联系人 查找联系人 修改联系人 清空联系人 退出功能 ...
C++核心_1内存分区模型
qq_45972323的博客
02-07 958
C++核心_1内存分区模型一级目录二级目录三级目录 一级目录 二级目录 三级目录
C++入门_8结构体
qq_45972323的博客
01-28 843
C++笔记_8结构体结构体结构体概念结构体定义和使用结构体数组结构体指针结构体嵌套结构体结构体做函数参数结构体中const使用场景结构体案例 结构体 结构体概念 结构体定义和使用 结构体数组 结构体指针 结构体嵌套结构体 结构体做函数参数 结构体中const使用场景 结构体案例 ...
C++入门_2数据类型
qq_45972323的博客
01-24 687
C++笔记_2数据类型数据类型与sizeof关键字布尔类型整形浮点型字符型转义字符字符串数据输入 数据类型与sizeof关键字 布尔类型 整形 浮点型 字符型 转义字符 字符串 数据输入
o2o优惠券使用预测
08-31
您好！针对O2O优惠券使用预测的问题，您可以考虑使用机器学习算法来进行预测。以下是一个简单的预测流程： 1. 数据收集：收集O2O优惠券使用的历史数据，包括用户的消费行为、领取优惠券的情况以及优惠券的属性等信息。 2. 数据预处理：对收集到的数据进行清洗和处理，包括处理缺失值、异常值和重复值等，同时进行特征工程，从原始数据中提取有用的特征。 3. 特征选择：根据问题的具体情况，选择合适的特征进行建模，可以考虑用户的历史消费情况、优惠券的属性、用户的个人信息等。 4. 模型选择和训练：选择适合的机器学习模型进行训练。常用的模型包括逻辑回归、决策树、随机森林、支持向量机等。可以使用交叉验证等技术来评估模型性能并选择最佳模型。 5. 模型评估和优化：使用测试集对训练好的模型进行评估，评估指标可以包括准确率、召回率、F1值等。如果模型表现不佳，可以考虑调整模型参数或者尝试其他模型。 6. 使用模型进行预测：使用训练好的模型对新数据进行预测，可以根据用户的消费行为、优惠券属性等信息，预测用户是否会使用优惠券。 需要注意的是，预测的准确性可能会受到数据质量和特征选择的影响，因此在进行预测之前，确保数据的质量和特征的有效性非常重要。另外，不同业务场景下可能需要调整模型和特征选择的方法，具体情况需要根据实际需求进行调整。希望这些信息对您有所帮助！如果还有其他问题，请随时提问。
评论
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