Land cover classification tutorial start issue

[heksam]

Hi!
Im trying out the land cover classification tutorial, but cant get past the first part of code. I have tried "pip install dask_cuda" in a terminal, but then this error pops up. I have tried other ways to install in terminal (and also in the notebook itself), but nothing works. Has anyone got a clue of what I could do? I dont understand why I need to install packages to run tutorials provided by planetary computer.

ImportError Traceback (most recent call last)
File /srv/conda/envs/notebook/lib/python3.11/site-packages/dask/array/init.py:4
3 try:
----> 4 from dask.array import backends, fft, lib, linalg, ma, overlap, random
5 from dask.array.blockwise import atop, blockwise

File /srv/conda/envs/notebook/lib/python3.11/site-packages/dask/array/fft.py:17
16 from dask.array.core import concatenate as _concatenate
---> 17 from dask.array.creation import arange as _arange
18 from dask.array.numpy_compat import NUMPY_GE_200

File /srv/conda/envs/notebook/lib/python3.11/site-packages/dask/array/creation.py:26
25 from dask.array.numpy_compat import AxisError
---> 26 from dask.array.ufunc import greater_equal, rint
27 from dask.array.utils import meta_from_array

File /srv/conda/envs/notebook/lib/python3.11/site-packages/dask/array/ufunc.py:10
9 from dask.array.core import Array, apply_infer_dtype, asarray, blockwise, elemwise
---> 10 from dask.base import is_dask_collection, normalize_function
11 from dask.highlevelgraph import HighLevelGraph

ImportError: cannot import name 'normalize_function' from 'dask.base' (/srv/conda/envs/notebook/lib/python3.11/site-packages/dask/base.py)

The above exception was the direct cause of the following exception:

ImportError Traceback (most recent call last)
File /srv/conda/envs/notebook/lib/python3.11/site-packages/dask/dataframe/init.py:98
97 from dask.base import compute
---> 98 from dask.dataframe import backends, dispatch, rolling
99 from dask.dataframe.core import (
100 DataFrame,
101 Index,
(...)
107 to_timedelta,
108 )

File /srv/conda/envs/notebook/lib/python3.11/site-packages/dask/dataframe/backends.py:10
8 from pandas.api.types import is_scalar, union_categoricals
---> 10 from dask.array.core import Array
11 from dask.array.dispatch import percentile_lookup

File /srv/conda/envs/notebook/lib/python3.11/site-packages/dask/array/init.py:273
267 msg = (
268 "Dask array requirements are not installed.\n\n"
269 "Please either conda or pip install as follows:\n\n"
270 " conda install dask # either conda install\n"
271 ' python -m pip install "dask[array]" --upgrade # or python -m pip install'
272 )
--> 273 raise ImportError(str(e) + "\n\n" + msg) from e

ImportError: cannot import name 'normalize_function' from 'dask.base' (/srv/conda/envs/notebook/lib/python3.11/site-packages/dask/base.py)

Dask array requirements are not installed.

Please either conda or pip install as follows:

conda install dask # either conda install
python -m pip install "dask[array]" --upgrade # or python -m pip install

The above exception was the direct cause of the following exception:

ImportError Traceback (most recent call last)
Cell In[2], line 2
1 from dask.distributed import Client
----> 2 from dask_cuda import LocalCUDACluster
4 cluster = LocalCUDACluster(threads_per_worker=4)
5 client = Client(cluster)

File /srv/conda/envs/notebook/lib/python3.11/site-packages/dask_cuda/init.py:8
6 import dask
7 import dask.utils
----> 8 import dask.dataframe.core
9 import dask.dataframe.shuffle
10 import dask.dataframe.multi

File /srv/conda/envs/notebook/lib/python3.11/site-packages/dask/dataframe/init.py:159
152 except ImportError as e:
153 msg = (
154 "Dask dataframe requirements are not installed.\n\n"
155 "Please either conda or pip install as follows:\n\n"
156 " conda install dask # either conda install\n"
157 ' python -m pip install "dask[dataframe]" --upgrade # or python -m pip install'
158 )
--> 159 raise ImportError(msg) from e
162 from dask.dataframe._testing import test_dataframe

ImportError: Dask dataframe requirements are not installed.

Please either conda or pip install as follows:

conda install dask # either conda install
python -m pip install "dask[dataframe]" --upgrade # or python -m pip install

[TomAugspurger]

Mmm, looks like I might have dropped dask-cuda when I did the update to resolve #321.

In the meantime, I wonder if can work around it with something like this (untested):

client = Client(threads_per_worker=4, n_workers=1)

instead of importing and using LocalCUDACluster. That'll lose some of the features dask-cuda brings, but I think we aren't using any of those (the GPU code will still run on the GPU).

[heksam]

It got me a little further, but now it stopped on the "Now let’s do some visual spot checking of our model." part in the middle.

import matplotlib.colors
from bokeh.models.tools import BoxZoomTool
import panel
import hvplot.xarray # noqa
etc

Here it throws a long error, bottom message saying "TypeError: argument of type 'ListedColormap' is not iterable"