create a mi

In [13]: mi = pd.MultiIndex.from_product([[1,2,3], list('abc'), pd.date_range('20200101', periods=2, tz='UTC')], names=['int', 'string', 'dt'])                            

You can already get .dtypes, but is slightly cumbersome. I would propose adding Multidex.dtypes (we already have MultiIndex.dtype but its always object). I think this is worth the convenience api.

I just ran into an issue when trying to use to_csv with distributed workers that don't share a file system. I shouldn't have been surprised that writing to a local file system from a distributed worker doesn't work. It shouldn't work. But the error I got was just a File Not Found error. That brought me to:dask/dask#2656 (comment) - which was the answer.

The current version of the HANS dataset is missing the additional information provided for each example, including the sentence parses, heuristic and subcase.

Current default value for rows_per_chunk parameter of the CSV writer is 8, which means that the input table is by default broken into many small slices that are written out sequentially. This reduces the performance by an order on magnitude in some cases.

In Python layer, the default is the number of rows (i.e. write table out in a single pass). We can follow this by setting rows_per_chunk

Support Series.median()

MCVE Code Sample

# Your code here
import numpy as np
import xarray as xr

data = np.zeros((10, 4))
example_xr = xr.DataArray(data, coords=[range(10), ["