为所有时间维度创建列

Question

我有像下面的 Data_df 示例数据框这样的数据。我想知道是否有一种方法可以为某个时间戳字段中的所有时间维度创建新列，例如“start_timestamp”字段。我想根据“start_timestamp”列为年、月、工作日、小时、分钟创建新列。我知道我可以手动为每个时间维度编码，但我想知道是否有办法检查时间戳并自动创建它们。

数据_df：

   Unnamed: 0  call_history_id                            calllog_id  \
0       16358       1210746736  ca58d850-6fe6-4673-a049-ea4a2d8d7ecf   
1       16361       1210976828  c005329b-955d-4d88-98a5-1c47e6a1cb80   
2       16402       1217791595  050e9b83-54c2-4c87-abdd-32225c0d3189   
3       16471       1228495414  45705ed1-a8e2-4a15-8941-5b0a40b7d409   
4       27906       1245173592  04e56818-04a0-4704-ac86-31c31dac2370   

        call_id  connection_id  pbx_name    pbx_id  extension_number  \
0  1.509170e+12   1.509170e+12  sales8x8  sales8x8               595   
1  1.509170e+12   1.509170e+12  sales8x8  sales8x8               595   
2  1.509170e+12   1.509170e+12  sales8x8  sales8x8               595   
3  1.509170e+12   1.509170e+12  sales8x8  sales8x8               595   
4  1.509170e+12   1.509170e+12  sales8x8  sales8x8               595   

  extension_id   customer_id      address                 name  \
0          595  2.525100e+29  14086694428           Sun Basket   
1          595  2.525100e+29  13214371589          PEREZ,BRYAN   
2          595  2.525100e+29  14088566290          14088566290   
3          595  2.525100e+29   8059316676              Dialing   
4          595  2.525100e+29  12028071151  Implementation Team   

  start_timestamp direction  call_internal  call_missed  duration  \
0    1/8/18 19:49         I            0.0          0.0    4414.0   
1    1/8/18 20:09         I            0.0          0.0    8300.0   
2    1/9/18 20:31         I            0.0          0.0   14766.0   
3   1/11/18 17:16         I            0.0          0.0    1686.0   
4   1/15/18 22:55         I            0.0          0.0    3491.0   

  device_model  group_call group_name group_number           device_id  \
0  mediaserver         0.0          N            N  MasterSlaveService   
1  mediaserver         0.0          N            N  MasterSlaveService   
2  mediaserver         0.0          N            N  MasterSlaveService   
3  mediaserver         0.0          N            N  MasterSlaveService   
4  mediaserver         0.0          N            N  MasterSlaveService   

  history_event_state   created_time   updated_time group_type  
0                   A   1/8/18 19:49   1/8/18 19:49          N  
1                   A   1/8/18 20:09   1/8/18 20:09        NaN  
2                   A   1/9/18 20:31   1/9/18 20:31          N  
3                   A  1/11/18 17:16  1/11/18 17:16          N  
4                   A  1/15/18 22:55  1/15/18 22:55          N  

更新：

def ts_periods(f_nm, d_list, d_df):
    t_df=d_df.copy()

    for i in d_list:
        if i=='year':
            t_df[f_nm+'_Year']=pd.DatetimeIndex(t_df[f_nm]).year
        elif i=='month':
            t_df[f_nm+'_month']=pd.DatetimeIndex(t_df[f_nm]).month
        elif i=='weekday':
            t_df[f_nm+'_weekday']=pd.DatetimeIndex(t_df[f_nm]).weekday_name
        elif i=='week' in d_list:
            t_df[f_nm+'_week']=pd.DatetimeIndex(t_df[f_nm]).week
        elif i=='hour':
            t_df[f_nm+'_hour']=pd.DatetimeIndex(t_df[f_nm]).hour
        elif i=='minute':
            t_df[f_nm+'_minute']=pd.DatetimeIndex(t_df[f_nm]).minute
    return t_df

Answer 1

使用您的数据和.dt 访问器的简短示例。我们首先将数据转换为 pandas 时间戳，然后访问我们想要的维度：

import pandas as pd

data = pd.DataFrame(
    {
        'time_stamp': ['1/8/18 19:49', '1/9/18 20:31', '1/11/18 17:16']
    }
)

data['time_stamp'] = pd.to_datetime(data['time_stamp'], dayfirst = True)

data['day_of_week'] = data['time_stamp'].dt.weekday

data['hour_of_day'] = data['time_stamp'].dt.hour

print(data)

给予：

           time_stamp  day_of_week  hour_of_day
0 2018-08-01 19:49:00            2           19
1 2018-09-01 20:31:00            5           20
2 2018-11-01 17:16:00            3           17

文档：https://pandas.pydata.org/pandas-docs/stable/basics.html#basics-dt-accessors