[BUG] RandomForestClassifier score failing due to output_class handling

[beckernick]

In today's nightly, RandomForestClassifier.score is failing with an output class handling error.

import cuml
from sklearn.datasets import make_classification
​
X, y = make_classification(
    n_samples=100000,
    n_features=20,
    n_informative=15,
    n_classes=2,
    class_sep=0.5,
    shuffle=True,
    random_state=12,
)
​
X = X.astype("float32")
​
gclf = cuml.ensemble.RandomForestClassifier(
    n_estimators=100,
)
gclf.fit(X, y)
gclf.score(X, y)
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-30-1162551a0613> in <module>
     18 )
     19 gclf.fit(X, y)
---> 20 gclf.score(X, y)

/raid/nicholasb/miniconda3/envs/rapids-gpubdb-20210331/lib/python3.7/site-packages/cuml/internals/api_decorators.py in inner(*args, **kwargs)
    358         def inner(*args, **kwargs):
    359             with self._recreate_cm(func, args):
--> 360                 return func(*args, **kwargs)
    361 
    362         return inner

cuml/ensemble/randomforestclassifier.pyx in cuml.ensemble.randomforestclassifier.RandomForestClassifier.score()

/raid/nicholasb/miniconda3/envs/rapids-gpubdb-20210331/lib/python3.7/site-packages/cuml/internals/api_decorators.py in inner_get(*args, **kwargs)
    584 
    585                 # Call the function
--> 586                 ret_val = func(*args, **kwargs)
    587 
    588             return cm.process_return(ret_val)

cuml/ensemble/randomforestclassifier.pyx in cuml.ensemble.randomforestclassifier.RandomForestClassifier.predict()

TypeError: predict() got an unexpected keyword argument 'output_class'

conda list | grep "rapids\|dask\|numpy\|cupy"
# packages in environment at /raid/nicholasb/miniconda3/envs/rapids-gpubdb-20210331:
cudf                      0.19.0a210331   cuda_10.2_py37_gc99fcef41b_313    rapidsai-nightly
cuml                      0.19.0a210331   cuda10.2_py37_g83168076c_138    rapidsai-nightly
cupy                      8.6.0            py37h7fc54ca_0    conda-forge
dask                      2021.3.1           pyhd8ed1ab_0    conda-forge
dask-core                 2021.3.1           pyhd8ed1ab_0    conda-forge
dask-cuda                 0.19.0a210331           py37_45    rapidsai-nightly
dask-cudf                 0.19.0a210331   py37_gc99fcef41b_313    rapidsai-nightly
libcudf                   0.19.0a210331   cuda10.2_gbe2f0c000f_314    rapidsai-nightly
libcuml                   0.19.0a210331   cuda10.2_g83168076c_138    rapidsai-nightly
libcumlprims              0.19.0a210316   cuda10.2_ge7e82a0_12    rapidsai-nightly
librmm                    0.19.0a210331   cuda10.2_g9d1ba02_50    rapidsai-nightly
numpy                     1.19.5           py37haa41c4c_1    conda-forge
rmm                       0.19.0a210331   cuda_10.2_py37_g9d1ba02_50    rapidsai-nightly
ucx                       1.9.0+gcd9efd3       cuda10.2_0    rapidsai-nightly
ucx-proc                  1.0.0                       gpu    rapidsai-nightly
ucx-py                    0.19.0a210331   py37_gcd9efd3_46    rapidsai-nightly

[hcho3]

The bug is most likely introduced in #3609. I will submit a pull request today.

[hcho3]

I filed #3685 to restore the functionality of score(). Thanks!