首先定義幾個時間變數datetime、Timestamp、numpy.datetime64：

import datetime
import numpy as np
import pandas as pd
dt = datetime.datetime(2018, 9, 1)
# A strange way to extract a Timestamp object, there's surely a better way?
ts = pd.DatetimeIndex([dt])[0]
dt64 = np.datetime64(dt)

dt
Out[67]: datetime.datetime(2018
 
, 9, 1, 0, 0)

ts
Out[68]: Timestamp('2018-09-01 00:00:00')

dt64
Out[69]: numpy.datetime64('2018-09-01T00:00:00.000000')

轉換numpy.datetime64物件

1、直接用astype的方法進行轉化

dt64.astype(datetime.datetime)
Out[74]: datetime.datetime(2018, 9, 1, 0, 0)

2、先轉化為timestamp，再轉化為datetime

• 先轉化為timestamp：

(dt64-np.datetime64('1970-01-01T00:00:00Z'
 
)) / np.timedelta64(1, 's')
Out[72]: 1535760000.0

• 再將timestamp轉為datetime格式

datetime.datetime.fromtimestamp((dt64-np.datetime64('1970-01-01T00:00:00Z')) / np.timedelta64(1, 's'))
Out[76]: datetime.datetime(2018, 9, 1, 8, 0)