support pandas 2.1

[celestinoxp]

2023-09-04 19:59:23,262:WARNING:C:\Users\username\anaconda3\lib\site-packages\xgboost\data.py:338: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead
is_categorical_dtype(dtype) or is_pa_ext_categorical_dtype(dtype)

using:
windows x64
python 3.10
pycaret 3.0.4
pandas 2.1
xgboost snapshot (xgboost-2.1.0.dev0+adea842c838095b92bda2dde962a28a606513e57-py3-none-win_amd64.whl)

[celestinoxp]

more warnings:
2023-09-06 09:12:21,935:WARNING:C:\Users\celes\anaconda3\lib\site-packages\xgboost\data.py:335: FutureWarning: is_sparse is deprecated and will be removed in a future version. Check isinstance(dtype, pd.SparseDtype) instead.
if is_sparse(dtype):

2023-09-06 09:12:21,935:WARNING:C:\Users\celes\anaconda3\lib\site-packages\xgboost\data.py:338: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead
is_categorical_dtype(dtype) or is_pa_ext_categorical_dtype(dtype)

2023-09-06 09:12:21,989:WARNING:C:\Users\celes\anaconda3\lib\site-packages\xgboost\data.py:384: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead
if is_categorical_dtype(dtype):

2023-09-06 09:12:21,989:WARNING:C:\Users\celes\anaconda3\lib\site-packages\xgboost\data.py:359: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead
return is_int or is_bool or is_float or is_categorical_dtype(dtype)

2023-09-06 09:12:27,026:WARNING:C:\Users\celes\anaconda3\lib\site-packages\xgboost\data.py:335: FutureWarning: is_sparse is deprecated and will be removed in a future version. Check isinstance(dtype, pd.SparseDtype) instead.
if is_sparse(dtype):

2023-09-06 09:12:27,026:WARNING:C:\Users\celes\anaconda3\lib\site-packages\xgboost\data.py:338: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead
is_categorical_dtype(dtype) or is_pa_ext_categorical_dtype(dtype)

2023-09-06 09:12:27,080:WARNING:C:\Users\celes\anaconda3\lib\site-packages\xgboost\data.py:384: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead
if is_categorical_dtype(dtype):

2023-09-06 09:12:27,080:WARNING:C:\Users\celes\anaconda3\lib\site-packages\xgboost\data.py:359: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead
return is_int or is_bool or is_float or is_categorical_dtype(dtype)

2023-09-06 09:12:27,141:WARNING:C:\Users\celes\anaconda3\lib\site-packages\xgboost\data.py:520: FutureWarning: is_sparse is deprecated and will be removed in a future version. Check isinstance(dtype, pd.SparseDtype) instead.
if is_sparse(data):

[watson-sam]

Nice - thanks for this

[FlorinAndrei]

@trivialfis Looks like this fix didn't make it into 2.0.0. Is it going to be included in the next patch release?

[trivialfis]

It will be part of next major release 2.1. in the meantime, you can use nightly build of xgboost.

[FlorinAndrei]

@trivialfis Downgrading from pandas-2.1.1 to pandas-2.0.3 increases the time by 70% in some workloads where pandas is not even the main time consumer. It's much slower.

The xgboost nightly builds master wheels page shows nothing. It's just an empty page.

I'm stuck with a very verbose series of xgboost warnings. :(

[trivialfis]

The xgboost nightly builds master wheels page shows nothing. It's just an empty page.

cc @hcho3 I can confirm that the master page keeps loading without showing the installation links. Release branches, however, seem to be fine.

[hcho3]

Yeah I notice this too. Let me fix it.

[hcho3]

@FlorinAndrei The nightly build index page is now fixed: https://s3-us-west-2.amazonaws.com/xgboost-nightly-builds/list.html?prefix=master/.

@trivialfis The page errored out because there were too many nightly build files and I had to prune old files to restore the functionality. I created the lifecycle policy so that files older than 30 days would be automatically deleted from the S3 bucket.

[trivialfis]

Thank you for the quick fix!