According to the xarray documentation: xarray.DataArray is xarray’s implementation of a labeled, multi-dimensional array. To me, this means a DataArray is just numpy with labels. However, xarray provides a second data structure, called Dataset, whose documentation is xarray.Dataset is xarray’s multi-dimensional equivalent of a DataFrame. Since a Pandas DataFrame is just a 2D numpy array, with labels, can't I still use a DataArray instead? In fact, two DataArray like import numpy as np import xarray as xr foo = xr.DataArray( data=np.zeros((4, 3)), coords={ "x": np.arange(4), "y": np.arange(3), }) bar = xr.DataArray( data=np.zeros((4, 3)), coords={ "x": np.arange(4), "y": np.arange(3), }) can be grouped together using a string coordinate da = xr.DataArray( data=np.zeros((4, 3, 2)), coords={ "x": np.arange(4), "y": np.arange(3), "key": ["foo", "bar"], }) The second part of the Dataset documentation is a little clearer [a Dataset] is a dict-like container of labeled arrays (DataArray objects) with aligned dimensions. However, from the examples in the documentation it appears that some of the Dataset usage can be perfectly covered by a DataArray. Is there any other difference that I am missing? Continue reading...
