在python中获取具有时移的列的最近值

Question

我有一个以日期时间（以毫秒为单位）作为索引和价格列的数据框。我想创建一个新列，其价格与前 20 行最接近。例如：

Data Hora
2024-02-01 10:03:39.483   1.880
2024-02-01 10:03:40.540   1.900
2024-02-01 10:03:41.550   1.880
2024-02-01 10:03:43.563   1.890
2024-02-01 10:03:45.567   1.870
2024-02-01 10:03:45.583   1.890
2024-02-01 10:03:46.590   1.900
2024-02-01 10:03:48.620   1.930
2024-02-01 10:03:50.627   1.880
2024-02-01 10:03:51.630   1.890
2024-02-01 10:03:53.647   1.880
2024-02-01 10:03:55.753   1.900
2024-02-01 10:03:59.367   1.890
2024-02-01 10:04:02.497   1.910
2024-02-01 10:04:04.543   1.890
2024-02-01 10:04:05.550   1.860
2024-02-01 10:04:07.577   1.840
2024-02-01 10:04:08.157   1.850
2024-02-01 10:04:10.197   1.880
2024-02-01 10:04:11.887   1.910
2024-02-01 10:04:13.163   1.920

Answer 1

假设您想获得最近 20 秒内最接近的价格，您可以使用

rolling.apply

:

df['Data Hora'] = pd.to_datetime(df['Data Hora'])

def closest(s):
    if len(s)>1:
        diff = s.iloc[:-1].sub(s.iloc[-1]).abs()
        return s.loc[diff.idxmin()]
    else:
        return float('nan')
    
df['closest_price_20s'] = df.rolling('20s', on='Data Hora')['Price'].apply(closest)

输出：

                 Data Hora  Price  closest_price_20s
0  2024-02-01 10:03:39.483   1.88                NaN
1  2024-02-01 10:03:40.540   1.90               1.88
2  2024-02-01 10:03:41.550   1.88               1.88
3  2024-02-01 10:03:43.563   1.89               1.88
4  2024-02-01 10:03:45.567   1.87               1.88
5  2024-02-01 10:03:45.583   1.89               1.89
6  2024-02-01 10:03:46.590   1.90               1.90
7  2024-02-01 10:03:48.620   1.93               1.90
8  2024-02-01 10:03:50.627   1.88               1.88
9  2024-02-01 10:03:51.630   1.89               1.89
10 2024-02-01 10:03:53.647   1.88               1.88
11 2024-02-01 10:03:55.753   1.90               1.90
12 2024-02-01 10:03:59.367   1.89               1.89
13 2024-02-01 10:04:02.497   1.91               1.90
14 2024-02-01 10:04:04.543   1.89               1.89
15 2024-02-01 10:04:05.550   1.86               1.87
16 2024-02-01 10:04:07.577   1.84               1.86
17 2024-02-01 10:04:08.157   1.85               1.86
18 2024-02-01 10:04:10.197   1.88               1.88
19 2024-02-01 10:04:11.887   1.91               1.91
20 2024-02-01 10:04:13.163   1.92               1.91

图形输出：