python - 从 datetime64[ns, UTC] 中提取年月日，Python

Question

我在 df 中有这个专栏：

    > df["time"]
0         2007-02-01 22:00:00+00:00
1         2007-02-01 22:00:00+00:00
2         2007-02-01 22:00:00+00:00
3         2007-02-01 22:00:00+00:00
4         2007-02-01 22:00:00+00:00

我想创建三个包含日、月和年的新列，但我想不出一种方法来提取time column.

标签： pythonpandasdatetimetimedatetime64

Answer 1

为了不修改现有time列，请使用pd.to_datetime然后使用dt访问器创建一个单独的日期时间序列：

# obtain datetime series:
datetimes = pd.to_datetime(df['time'])

# assign your new columns
df['day'] = datetimes.dt.day
df['month'] = datetimes.dt.month
df['year'] = datetimes.dt.year

>>> df
                        time  day  month  year
0  2007-02-01 22:00:00+00:00    1      2  2007
1  2007-02-01 22:00:00+00:00    1      2  2007
2  2007-02-01 22:00:00+00:00    1      2  2007
3  2007-02-01 22:00:00+00:00    1      2  2007
4  2007-02-01 22:00:00+00:00    1      2  2007

另一种方法是str.split('-')在datetime.dt.date系列上使用：

datetimes = pd.to_datetime(df['time'])

df[['year','month','day']] = datetimes.dt.date.astype(str).str.split('-',expand=True)

>>> df
                        time  year month day
0  2007-02-01 22:00:00+00:00  2007    02  01
1  2007-02-01 22:00:00+00:00  2007    02  01
2  2007-02-01 22:00:00+00:00  2007    02  01
3  2007-02-01 22:00:00+00:00  2007    02  01
4  2007-02-01 22:00:00+00:00  2007    02  01