将pandas datetime列舍入到前一分钟

Question

我的数据框如下：

      price                timestamp  amount  side
0  0.001728  2019-05-01 22:48:49.082    29.0   buy
1  0.001729  2019-05-01 21:26:22.078    31.0  sell
2  0.001730  2019-05-01 18:05:30.953    32.0   buy
3  0.001726  2019-05-01 18:05:40.412    32.0  sell
4  0.001730  2019-05-01 16:59:51.127    32.0  sell
5  0.001741  2019-05-01 12:23:19.957    32.0   buy

我想要的是以下内容：

      price           timestamp  amount  side
0  0.001728 2019-05-01 22:48:00    29.0   buy
1  0.001729 2019-05-01 21:26:00    31.0  sell
2  0.001730 2019-05-01 18:05:00    32.0   buy
3  0.001726 2019-05-01 18:05:00    32.0  sell
4  0.001730 2019-05-01 16:59:00    32.0  sell
5  0.001741 2019-05-01 12:23:00    32.0   buy

我尝试过的：

df['timestamp'] = pd.to_datetime(df['timestamp'], unit='%Y%m%d %H:%M')

但是它并没有改变我的数据框...为什么知道呢？谢谢！

Answer 1

尝试使用dt.floor进行前一分钟的发言：

pd.to_datetime(df['timestamp'], errors='coerce').dt.floor('min')

0   2019-05-01 22:48:00
1   2019-05-01 21:26:00
2   2019-05-01 18:05:00
3   2019-05-01 18:05:00
4   2019-05-01 16:59:00
5   2019-05-01 12:23:00
Name: timestamp, dtype: datetime64[ns]

---

df.assign(
    timestamp=pd.to_datetime(df['timestamp'], errors='coerce').dt.floor('min'))

      price           timestamp  amount  side
0  0.001728 2019-05-01 22:48:00    29.0   buy
1  0.001729 2019-05-01 21:26:00    31.0  sell
2  0.001730 2019-05-01 18:05:00    32.0   buy
3  0.001726 2019-05-01 18:05:00    32.0  sell
4  0.001730 2019-05-01 16:59:00    32.0  sell
5  0.001741 2019-05-01 12:23:00    32.0   buy

Answer 2

Numpy的astype

因为@ cs95可以让我更快地给出答案（-：

df.assign(timestamp=df.timestamp.astype('datetime64[m]'))

      price           timestamp  amount  side
0  0.001728 2019-05-01 22:48:00    29.0   buy
1  0.001729 2019-05-01 21:26:00    31.0  sell
2  0.001730 2019-05-01 18:05:00    32.0   buy
3  0.001726 2019-05-01 18:05:00    32.0  sell
4  0.001730 2019-05-01 16:59:00    32.0  sell
5  0.001741 2019-05-01 12:23:00    32.0   buy