在 Python 中重新采样和插值时间序列

Question

我有带有风向 (wd) 和速度 (ws) 的 csv 文件：

datetime  wd  ws
06.02.2023 00:55  297  3.2
06.02.2023 01:55  296  2.7
06.02.2023 02:55  299  3.0
06.02.2023 03:55  302  3.5

我想将时间重新采样为一小时，如下所示：

datetime  wd  ws
06.02.2023 01:00  297  3.2
06.02.2023 02:00  296  2.7
06.02.2023 03:00  299  3.0
06.02.2023 04:00  302  3.5

到目前为止，我一直在尝试使用这个脚本：

import pandas as pd
filename = r"data.csv"
df = pd.read_csv(filename, header=None, sep=";", skiprows=1, usecols = [0,3,4], names = ["datetime","wd","ws"])

# sspecifying the date format
df['index_time']= pd.to_datetime(df['local_time'], format='%d.%m.%Y %H:%M')

# change date-time column to index
df.set_index('index_time', inplace=True)

# trying to resample
df_resampled = df.resample(rule='H')

打印（df_resampled）输出：

                           local_time     wd   ws
index_time                                       
2023-02-06 00:00:00               NaN    NaN  NaN
2023-02-06 01:00:00               NaN    NaN  NaN
2023-02-06 02:00:00               NaN    NaN  NaN
2023-02-06 03:00:00               NaN    NaN  NaN

如何仅对时间进行重新采样而保留数据不变？

Answer 1

也许我误解了，但您似乎只是想将时间向上或向下四舍五入到最接近的小时？如果是这样，那么以下代码应该可以工作：

import pandas as pd
from datetime import datetime, timedelta

data = {
    'date': ['06.02.2023', '06.02.2023', '06.02.2023', '06.02.2023'],
    'time': ['0:55', '1:55', '2:55', '3:55'],
    'wd': [297, 296, 299, 302],
    'ws': [3.2, 2.7, 3.0, 3.5]
}

df = pd.DataFrame(data)

def round_up_to_nearest_hour(time_str):
    time_obj = datetime.strptime(time_str, '%H:%M') #Convert string of time to time object
    if time_obj.minute >= 30: #If minutes are above 30...
        time_obj += timedelta(hours=1) #...increase the hour by 1
    time_obj = time_obj.replace(minute=0) #Reset minutes to zero
    return time_obj.strftime('%H:%M') #Return the new time

df['time'] = df['time'].apply(round_up_to_nearest_hour) #Apply this function to all values in the time column of df

我在 Jupyter 中运行了这个（见下文：我将一个值切换到低于 30 分钟标记），看起来它正在做你想要的事情？希望这有帮助！