不可哈希类型：将 modin 与 pandas 一起使用时的系列？

Question

我在 Windows 10 上使用 Anaconda；我通过以下方式安装：conda install -c anaconda daskconda install -c conda-forge modinconda update condaconda update anacondaconda update daskconda install -c conda-forge pandas=1.0.5  # this will also download modin 0.7.4-py_0 --> 0.8.0-py_0因此，请考虑以下示例：#!/usr/bin/env python3import ioUSEDASK=Falseif not USEDASK:  import pandas as pdelse:  from dask.distributed import Client # SO:48067066  client = Client(processes=False)  # create scheduler and worker automatically  #os.environ["MODIN_ENGINE"] = "dask"  # Modin will use Dask  import modin.pandas as pdmy_csv_str = """Time[s], Channel 00.000000000000000, -0.7366808056831360.000008000000000, -0.7264851927757260.000016000000000, -0.7213873863220210.000024000000000, -0.7111917734146120.000032000000000, -0.7009961605072020.000040000000000, -0.6908005475997920.000048000000000, -0.6704093217849730.000056000000000, -0.655115902423859"""my_csv_io = io.StringIO()my_csv_io.write(my_csv_str)my_csv_io.seek(0)my_df = pd.read_csv(my_csv_io)my_df.index = pd.to_timedelta(my_df.iloc[:,0], unit='s')print(my_df)当我拥有时USEDASK=False，一切都会按预期进行。当我有 时USEDASK=True，我遇到以下失败：python test\test.pyUserWarning: The Dask Engine for Modin is experimental.UserWarning: Parameters provided defaulting to pandas implementation.To request implementation, send an email to [e-mail] (most recent call last):  File "test\test.py", line 30, in <module>    my_df.index = pd.to_timedelta(my_df.iloc[:,0], unit='s')  File "C:\ProgramData\Anaconda3\lib\site-packages\pandas\core\tools\timedeltas.py", line 102, in to_timedelta    return _convert_listlike(arg, unit=unit, errors=errors)有什么方法可以让这段代码使用 modin+dask 工作吗？

Answer 1

这不是一个好的解决方案，但应该使用解决方法：

my_df.index = pd.to_timedelta(my_df.iloc[:,0].values, unit='s')

这对两者都适用USEDASK是对还是错