数据帧追加和drop_duplicates问题

Question

所以，我有一个这样的虚拟df，并将其保存到csv中：

import pandas as pd
import io

old_data = """date,time,open,high,low,close,volume
2021-05-06,04:08:00,9150090.0,9150090.0,9125001.0,9130000.0,9.015642
2021-05-06,04:09:00,9140000.0,9145000.0,9125012.0,9134068.0,3.121043
2021-05-06,04:10:00,9133882.0,9133882.0,9125002.0,9132999.0,5.536345
2021-05-06,04:11:00,9132999.0,9135013.0,9131000.0,9132999.0,5.880620"""

new_data = """timestamp,open,high,low,close,volume
1620274080000,9150090.0,9150090.0,9125001.0,9130000.0,9.015641820000004
1620274140000,9140000.0,9145000.0,9125012.0,9134068.0,3.121042509999999
1620274200000,9133882.0,9133882.0,9125002.0,9132999.0,5.5363449
1620274260000,9132999.0,9135013.0,9131000.0,9132999.0,5.88062024"""

我尝试检查df_old和df_new之间是否有重复的数据，如果有，我就丢弃它：

raw = pd.read_csv(io.StringIO(new_data), encoding='UTF-8')

stream = pd.DataFrame(raw, columns=['timestamp', 'open', 'high', 'low', 'close', 'volume'])
stream['timestamp'] = pd.to_datetime(stream['timestamp'], unit='ms')
stream['date'] = pd.to_datetime(stream['timestamp']).dt.date
stream['time'] = pd.to_datetime(stream['timestamp']).dt.time
stream = stream[['date', 'time', 'open', 'high', 'low', 'close', 'volume']]

for dif_date in stream.date.unique():
    grouped = stream.groupby(stream.date)
    df_new = grouped.get_group(dif_date)
    df_old = pd.read_csv(io.StringIO(old_data), encoding='UTF-8')

df_stream = df_old.append(df_new).reset_index(drop=True)
df_stream = df_stream.drop_duplicates(subset=['time'])
print(df_stream)

>    date        time      open       high       low        close      volume
> 0  2021-05-06  04:08:00  9150090.0  9150090.0  9125001.0  9130000.0  9.015642
> 1  2021-05-06  04:09:00  9140000.0  9145000.0  9125012.0  9134068.0  3.121043
> 2  2021-05-06  04:10:00  9133882.0  9133882.0  9125002.0  9132999.0  5.536345
> 3  2021-05-06  04:11:00  9132999.0  9135013.0  9131000.0  9132999.0  5.880620
> 4  2021-05-06  04:08:00  9150090.0  9150090.0  9125001.0  9130000.0  9.015642
> 5  2021-05-06  04:09:00  9140000.0  9145000.0  9125012.0  9134068.0  3.121043
> 6  2021-05-06  04:10:00  9133882.0  9133882.0  9125002.0  9132999.0  5.536345
> 7  2021-05-06  04:11:00  9132999.0  9135013.0  9131000.0  9132999.0  5.880620

但是结果还是返回了重复的值，如何解决这个问题或者重新排序？https://colab.research.google.com/drive/1vMx9hXKcbz8SDawTnHbzpV6JiRZsEuVP?usp=sharing之前谢谢

Answer 1

沿time列的类型不是常量，因此python无法判断行是否相等。

例如，如果您运行：

df_stream.time.loc[0] == df_stream.time.loc[4]

您将得到False，因为左侧是一个字符串，右侧是一个datetime.time对象。

应使用astype()在列'time‘上强制输入类型