import pandas as pd
lists=pd.DataFrame({'a':[1,2,3,4,5,6,7,8,9]})
lists['b']=(lists['a']<6).mul(lists['a']>4)

df["new"]=df3["rate"]*df3["duration"]

#將LBL1特征的值衍生為one-hot形式的新特征
piao=df_train_log.LBL1.value_counts().index
#先構(gòu)造一個(gè)臨時(shí)的df
df_tmp=pd.DataFrame({'USRID':df_train_log.drop_duplicates('USRID').USRID.values})
#將所有的新特征列都置為0
for i in piao:
 df_tmp['PIAO_'+i]=0
#進(jìn)行分組便利，有這個(gè)特征就置為1，原數(shù)據(jù)每個(gè)USRID有多條記錄，所以分組統(tǒng)計(jì)
group=df_train_log.groupby(['USRID'])
for k in group.groups.keys():
 t = group.get_group(k)
 id=t.USRID.value_counts().index[0]
 tmp_list=t.LBL1.value_counts().index
 for j in tmp_list:
  df_tmp['PIAO_'+j].loc[df_tmp.USRID==id]=1

group=df_train_log.groupby(['USRID'])
lt=[]
list_max_lbl1=[]
list_max_lbl2=[]
list_max_lbl3=[]
for k in group.groups.keys():
 t = group.get_group(k)
 #通過(guò)value_counts找出出現(xiàn)次數(shù)最多的項(xiàng)
 argmx = np.argmax(t['EVT_LBL'].value_counts())
 lbl1_max=np.argmax(t['LBL1'].value_counts())
 lbl2_max=np.argmax(t['LBL2'].value_counts())
 lbl3_max=np.argmax(t['LBL3'].value_counts())
 list_max_lbl1.append(lbl1_max)
 list_max_lbl2.append(lbl2_max)
 list_max_lbl3.append(lbl3_max)
 #只留下出現(xiàn)次數(shù)最多的項(xiàng)
 c = t[t['EVT_LBL']==argmx].drop_duplicates('EVT_LBL')
 #放入list中
 lt.append(c)
#構(gòu)造一個(gè)新的df
df_train_log_new = pd.concat(lt)
#另外又構(gòu)造了三個(gè)特征，LBL1-LBL3分別出現(xiàn)次數(shù)最多的項(xiàng)
df_train_log_new['LBL1_MAX']=list_max_lbl1
df_train_log_new['LBL2_MAX']=list_max_lbl2
df_train_log_new['LBL3_MAX']=list_max_lbl3

#創(chuàng)造臨時(shí)df，星期三，星期六，星期七，都默認(rèn)置為0
df_day=pd.DataFrame({'USRID':df_train_log.drop_duplicates('USRID').USRID.values})
df_day['weekday_3']=0
df_day['weekday_6']=0
df_day['weekday_7']=0
#分組統(tǒng)計(jì)，有就置為1，沒(méi)有置為0
group=df_train_log.groupby(['USRID'])
for k in group.groups.keys():
 t = group.get_group(k)
 id=t.USRID.value_counts().index[0]
 tmp_list=t.occ_dayofweek.value_counts().index
 for j in tmp_list:
  if j==3:
df_day['weekday_3'].loc[df_tmp.USRID==id]=1
  elif j==6:
df_day['weekday_6'].loc[df_tmp.USRID==id]=1
  elif j==7:
df_day['weekday_7'].loc[df_tmp.USRID==id]=1

#首先將日期轉(zhuǎn)化為時(shí)間戳，并賦予一個(gè)新特征
tmp_list=[]
for i in df_train_log.OCC_TIM:
 d=datetime.datetime.strptime(str(i),"%Y-%m-%d %H:%M:%S")
 evt_time = time.mktime(d.timetuple())
 tmp_list.append(evt_time)
df_train_log['time']=tmp_list
#每下一行減去上一行，得到app停留時(shí)間
df_train_log['diff_time']=df_train_log.time-df_train_log.time.shift(1)
#構(gòu)造一個(gè)新的dataFrame，分組得到查看app的天數(shù)
df_time=pd.DataFrame({'USRID':df_train_log.drop_duplicates('USRID').USRID.values})
#有幾天查看
df_time['days']=0
group=df_train_log.groupby(['USRID'])
for k in group.groups.keys():
 t = group.get_group(k)
 id=set(t.USRID).pop()
 df_time['days'].loc[df_time.USRID==id]= len(t.occ_day.value_counts().index)
#去掉一些異常時(shí)間戳，比如間隔兩天的相減，肯定不合適，na的也去掉了
df_train_log=df_train_log[(df_train_log.diff_time>0)&(df_train_log.diff_time<8000)]
#累計(jì)停留時(shí)間
group_stayTime=df_train_log['diff_time'].groupby(df_train_log['USRID']).sum()
#創(chuàng)造新的df
df_tmp=pd.DataFrame({'USRID':list(group_stayTime.index.values),'stay_time':list(group_stayTime.values)})
#合并成一個(gè)新的df
df=pd.merge(df_time,df_tmp,on=['USRID'],how='left')#合并后，缺失的停留時(shí)間，置為0df.fillna(0,axis=1,inplace=True)