如何插值然后在新的一天重新开始？

Question

继我之前的问题之后，我将如何做同样的事情，但在每天结束时停止。我试过分组，但这似乎删除了很多数据。这是我开始的数据：

time = np.array([pd.to_datetime("2022-01-01 00:00:00"),pd.to_datetime("2022-01-01 00:00:01"),pd.to_datetime("2022-01-01 00:00:03"), pd.to_datetime("2022-01-01 00:00:04"),pd.to_datetime("2022-01-02 00:00:07"),pd.to_datetime("2022-01-02 00:00:09"), pd.to_datetime("2022-01-02 00:00:10")])
lat = [58.1, 58.4, 58.5, 58.9, 52,52.2, 52.5]
lng = [1.34, 1.44, 1.46, 1.48, 1.35, 1.37, 1.39]

df =  pd.DataFrame({"time": time, "lat": lat, "lng" :lng})


time                lat     lng
2022-01-01 00:00:00 58.1    1.34
2022-01-01 00:00:01 58.4    1.44
2022-01-01 00:00:03 58.5    1.46
2022-01-01 00:00:04 58.9    1.48
2022-01-02 00:00:07 52.0    1.35
2022-01-02 00:00:09 52.2    1.37
2022-01-02 00:00:10 52.5    1.39

预期输出为：

time                lat     lng
2022-01-01 00:00:00 58.1    1.34
2022-01-01 00:00:01 58.4    1.44
2022-01-01 00:00:01 58.45   1.45
2022-01-01 00:00:03 58.5    1.46
2022-01-01 00:00:04 58.9    1.48
2022-01-02 00:00:07 52.0    1.35
2022-01-02 00:00:08 52.1    1.36
2022-01-02 00:00:09 52.2    1.37
2022-01-02 00:00:10 52.5    1.39

使用这个：

df = df.set_index('time').asfreq(freq='S').interpolate()

当我的所有数据都来自同一天时，效果很好。我怎样才能让它在第二天重置？

Answer 1

可以groupby and use a custom function with apply到运行相关的插值逻辑：

def func(x):
    return x.set_index('time').asfreq(freq='S').interpolate().reset_index()

df.groupby(df['time'].dt.day).apply(func).reset_index(drop=True)

结果：

                   time    lat   lng
0   2022-01-01 00:00:00  58.10  1.34
1   2022-01-01 00:00:01  58.40  1.44
2   2022-01-01 00:00:02  58.45  1.45
3   2022-01-01 00:00:03  58.50  1.46
4   2022-01-01 00:00:04  58.90  1.48
5   2022-01-02 00:00:07  52.00  1.35
6   2022-01-02 00:00:08  52.10  1.36
7   2022-01-02 00:00:09  52.20  1.37
8   2022-01-02 00:00:10  52.50  1.39

如何插值然后在新的一天重新开始？

How to interpolate and then start again on new day?

python

datetime

dataframe

pandas

imputation