Python 如何创建包含一天中部分时间（晚上、早上、下午、晚上）的新列 我希望创造一个包含一天中的一部分的新专栏。我会考虑{晚上（00∶01至6:00）、早晨（6:01至12:00）、下午（12:01至18:00）、晚上（18:011:00）}。 以下是数据帧： package_name name starttime duration UserId com.facebook.katana Facebook 2020-09-19 6:02:06.019 28.077 4 com.android.systemui System UI 2020-09-19 16:42:34.096 28.077 4 com.android.systemui System UI 2020-09-19 19:51:35.778 0.329 4 com.facebook.katana Facebook 2020-09-19 10:56:39.129 0.329 4 com.android.systemui System UI 2020-09-19 01:48:32.067 3.022 4

我在论坛上发现了类似的问题，但我无法修改代码以处理我的数据。

您可以执行以下操作：

In [4249]: df.starttime = pd.to_datetime(df.starttime)
In [4253]: conditions = [df.starttime.dt.hour.between(0, 6), df.starttime.dt.hour.between(6, 12), df.starttime.dt.hour.between(12, 18), df.starttime.dt.hour.between(18,24)]

In [4254]: choices = ['night', 'morning', 'afternoon', 'evening']

In [4257]: df['part_of_day'] = np.select(conditions, choices)

In [4258]: df
Out[4258]: 
           package_name       name               starttime  duration  UserId part_of_day
0   com.facebook.katana   Facebook 2020-09-19 06:02:06.019    28.077       4       night
1  com.android.systemui  System UI 2020-09-19 16:42:34.096    28.077       4   afternoon
2  com.android.systemui  System UI 2020-09-19 19:51:35.778     0.329       4     evening
3  com.android.systemui  System UI 2020-09-19 01:48:32.067     0.329       4       night

你可以做：

In [4249]: df.starttime = pd.to_datetime(df.starttime)
In [4253]: conditions = [df.starttime.dt.hour.between(0, 6), df.starttime.dt.hour.between(6, 12), df.starttime.dt.hour.between(12, 18), df.starttime.dt.hour.between(18,24)]

In [4254]: choices = ['night', 'morning', 'afternoon', 'evening']

In [4257]: df['part_of_day'] = np.select(conditions, choices)

In [4258]: df
Out[4258]: 
           package_name       name               starttime  duration  UserId part_of_day
0   com.facebook.katana   Facebook 2020-09-19 06:02:06.019    28.077       4       night
1  com.android.systemui  System UI 2020-09-19 16:42:34.096    28.077       4   afternoon
2  com.android.systemui  System UI 2020-09-19 19:51:35.778     0.329       4     evening
3  com.android.systemui  System UI 2020-09-19 01:48:32.067     0.329       4       night

如果性能很重要，请在此处使用将日期时间转换为小时：

df['starttime'] = pd.to_datetime(df['starttime'])

df['new'] = pd.cut(df['starttime'].dt.hour,
                   bins=[0,6,12,18,23],
                   labels=['night','morning','afternoon','evening'],
                   include_lowest=True)

print (df)
           package_name       name               starttime  duration  UserId  \
0   com.facebook.katana   Facebook 2020-09-19 06:02:06.019    28.077       4   
1  com.android.systemui  System UI 2020-09-19 16:42:34.096    28.077       4   
2  com.android.systemui  System UI 2020-09-19 19:51:35.778     0.329       4   
3   com.facebook.katana   Facebook 2020-09-19 10:56:39.129     0.329       4   
4  com.android.systemui  System UI 2020-09-19 01:48:32.067     3.022       4   

         new  
0      night  
1  afternoon  
2    evening  
3    morning  
4      night  

5k行的性能测试：

df['starttime'] = pd.to_datetime(df['starttime'])

df = pd.concat([df] * 1000, ignore_index=True)


In [79]: %%timeit
    ...: conditions = [df.starttime.dt.hour.between(0, 6), df.starttime.dt.hour.between(6, 12), df.starttime.dt.hour.between(12, 18), df.starttime.dt.hour.between(18,24)]
    ...: 
    ...: choices = ['night', 'morning', 'afternoon', 'evening']
    ...: 
    ...: df['part_of_day'] = np.select(conditions, choices)
    ...: 
5.28 ms ± 451 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

In [80]: %%timeit
    ...: df['new'] = pd.cut(df['starttime'].dt.hour,
    ...:                    bins=[0,6,12,18,23],
    ...:                    labels=['night','morning','afternoon','evening'],
    ...:                    include_lowest=True)
    ...:                    
2.1 ms ± 13.8 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

对于5万行：

df['starttime'] = pd.to_datetime(df['starttime'])

df = pd.concat([df] * 10000, ignore_index=True)


In [82]: %%timeit
    ...: conditions = [df.starttime.dt.hour.between(0, 6), df.starttime.dt.hour.between(6, 12), df.starttime.dt.hour.between(12, 18), df.starttime.dt.hour.between(18,24)]
    ...: 
    ...: choices = ['night', 'morning', 'afternoon', 'evening']
    ...: 
    ...: df['part_of_day'] = np.select(conditions, choices)
    ...: 
    ...: 
26.9 ms ± 221 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)

In [83]: %%timeit
    ...: df['new'] = pd.cut(df['starttime'].dt.hour,
    ...:                    bins=[0,6,12,18,23],
    ...:                    labels=['night','morning','afternoon','evening'],
    ...:                    include_lowest=True)
    ...:                    
7.46 ms ± 68.3 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

如果性能很重要，请在此处使用将日期时间转换为小时：

df['starttime'] = pd.to_datetime(df['starttime'])

df['new'] = pd.cut(df['starttime'].dt.hour,
                   bins=[0,6,12,18,23],
                   labels=['night','morning','afternoon','evening'],
                   include_lowest=True)

print (df)
           package_name       name               starttime  duration  UserId  \
0   com.facebook.katana   Facebook 2020-09-19 06:02:06.019    28.077       4   
1  com.android.systemui  System UI 2020-09-19 16:42:34.096    28.077       4   
2  com.android.systemui  System UI 2020-09-19 19:51:35.778     0.329       4   
3   com.facebook.katana   Facebook 2020-09-19 10:56:39.129     0.329       4   
4  com.android.systemui  System UI 2020-09-19 01:48:32.067     3.022       4   

         new  
0      night  
1  afternoon  
2    evening  
3    morning  
4      night  

5k行的性能测试：

df['starttime'] = pd.to_datetime(df['starttime'])

df = pd.concat([df] * 1000, ignore_index=True)


In [79]: %%timeit
    ...: conditions = [df.starttime.dt.hour.between(0, 6), df.starttime.dt.hour.between(6, 12), df.starttime.dt.hour.between(12, 18), df.starttime.dt.hour.between(18,24)]
    ...: 
    ...: choices = ['night', 'morning', 'afternoon', 'evening']
    ...: 
    ...: df['part_of_day'] = np.select(conditions, choices)
    ...: 
5.28 ms ± 451 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

In [80]: %%timeit
    ...: df['new'] = pd.cut(df['starttime'].dt.hour,
    ...:                    bins=[0,6,12,18,23],
    ...:                    labels=['night','morning','afternoon','evening'],
    ...:                    include_lowest=True)
    ...:                    
2.1 ms ± 13.8 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

对于5万行：

df['starttime'] = pd.to_datetime(df['starttime'])

df = pd.concat([df] * 10000, ignore_index=True)


In [82]: %%timeit
    ...: conditions = [df.starttime.dt.hour.between(0, 6), df.starttime.dt.hour.between(6, 12), df.starttime.dt.hour.between(12, 18), df.starttime.dt.hour.between(18,24)]
    ...: 
    ...: choices = ['night', 'morning', 'afternoon', 'evening']
    ...: 
    ...: df['part_of_day'] = np.select(conditions, choices)
    ...: 
    ...: 
26.9 ms ± 221 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)

In [83]: %%timeit
    ...: df['new'] = pd.cut(df['starttime'].dt.hour,
    ...:                    bins=[0,6,12,18,23],
    ...:                    labels=['night','morning','afternoon','evening'],
    ...:                    include_lowest=True)
    ...:                    
7.46 ms ± 68.3 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)