检查数据帧中某个时间段内的条件。[Python][Pandas]

Question

所以我有一个数据框，我想检查一下：

如果行为人和被害人在某一时间（时间戳）有亲属关系，在某一时间范围内，是否发生了相反的情况，被害人成为行为人，反之亦然。你知道吗

例如if X attacks Y at time Z如果Y在时间范围内攻击X（Z + 24hrs），则添加一个计数

        X       Perpetrator          Victim           Time Stamp
44      3              FvdP           Jmallios 2002-11-13 21:53:17   
45      2              FvdP               FvdP 2002-11-13 21:58:59   
46     28               Lir                Lir 2002-11-19 00:47:05   
47     16   194.117.133.118            Tinulim 2002-11-27 00:13:40   
48     15           Tinulim    194.117.133.118 2002-11-27 00:16:03   
49     14             Zocky            Tinulim 2002-11-27 00:23:28   
50     13   194.117.133.118              Zocky 2002-11-27 00:25:31   

这是数据帧的一部分。例如，用户194.117.133.118攻击用户Tinulim，然后用户在3分钟后194.117.133.118返回。[时间戳条目是datetime.Timestamp形式的对象]

我想添加两个用户和计数到一个新的数据帧。e、 g

         Initiator          Retort           Time Stamp
1  194.117.133.118         Tinulim  2002-11-27 00:13:40  

我当前正在将for循环与iterrows一起使用，但无法使用此函数正确格式化它：

def time_window_24(time1,time2):
    "Is time2 within 24 hours ahead of time1?"
    if (time2 - time1).days == 0 and time1!=time2:
        return True
    else:
        return False

Answer 1

这里有一个似乎可以使用pd.merge_asof的解决方案，您可以在这里看到API：merge_asof docs

data = """
perpetrator,victim,date
FvdP,Jmallios,2002-11-13 21:53:17
FvdP,FvdP,2002-11-13 21:58:59 
Lir,Lir,2002-11-19 00:47:05 
194.117.133.118,Tinulim,2002-11-27 00:13:40 
Tinulim,194.117.133.118,2002-11-27 00:16:03 
Zocky,Tinulim,2002-11-27 00:23:28 
194.117.133.118,Zocky,2002-11-27 00:25:31 
"""
from io import StringIO
df = pd.read_csv(StringIO(data), parse_dates=[2])
df = df.query('perpetrator != victim')
merged = pd.merge_asof(df,
                       df,
                       on=['date'],
                       right_by=['perpetrator', 'victim'],
                       left_by=['victim', 'perpetrator'],
                       direction='forward',
                       tolerance=pd.tslib.Timedelta('1 day'),
                       suffixes=('', '_res'))
merged = merged.rename(columns={'victim_res': 'initiator',
                                'perpetrator_res': 'retort'})
print(merged[merged.retort.notnull()][['initiator', 'retort', 'date']].to_string())

initiator retort date 1 194.117.133.118 Tinulim 2002-11-27 00:13:40