[BUG] SMOTENC with bool and pd.Categorical inputs

Describe the bug

SMOTENC fails to fit_resample if input columns of both bool and category types are present. In fact, it looks like the use of bool or of a categorical column with underlying bool data forces the conversion of other categorical input columns into float, which may end up in error if the conversion is not supported.

Steps/Code to Reproduce

import pandas as pd
from imblearn.over_sampling import SMOTENC

Xx = pd.DataFrame({
    'c':pd.Categorical([x for x in "abbacaba"*3]),
    'f':[.3,.5,.1,.2]*6,
    'b':[False, False, True]*8})
y = pd.DataFrame({'out':[1,0,0,0,0,1,0,0,1,1,0,0]*2})

# Xx.info(), y.info()

SMOTENC(categorical_features=[0]).fit_resample(Xx.drop(columns=['b']), y)                           # OK cat + float
SMOTENC(categorical_features=[0]).fit_resample(Xx.astype({'b':float}), y)                           # OK cat + float + float
SMOTENC(categorical_features=[0]).fit_resample(Xx.astype({'c':str}), y)                             # OK 2str/cat + float + bool
SMOTENC(categorical_features=[0,2]).fit_resample(Xx.astype({'b':str}), y)                           # OK cat + float + 2str/cat
SMOTENC(categorical_features=[0,2]).fit_resample(Xx.astype({'b':str}).astype({'b':'category'}), y)  # OK cat + float + 2str2cat/cat

SMOTENC(categorical_features=[0]).fit_resample(Xx, y)                                             # ERROR cat + float + bool
# SMOTENC(categorical_features=[0,2]).fit_resample(Xx, y)                                           # ERROR cat + float + bool/cat
# SMOTENC(categorical_features=[0,2]).fit_resample(Xx.astype({'b':'category'}), y)                  # ERROR cat + float + 2cat/cat

If column 'c' is defined from a string of digits instead ( e.g. 'c':pd.Categorical([x for x in "01102010"*3]) ) no exceptions are raised, but categorical columns in the output of the problematic calls present only Nan values

Expected Results

No error, balanced X,y in return

Actual Results

Traceback of the first faulty call in the example, other two are very similar

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
File ~/miniconda3/lib/python3.10/site-packages/pandas/core/arrays/categorical.py:551, in Categorical.astype(self, dtype, copy)
    550 try:
--> 551     new_cats = new_cats.astype(dtype=dtype, copy=copy)
    552     fill_value = self.categories._na_value

ValueError: could not convert string to float: 'a'

During handling of the above exception, another exception occurred:

ValueError                                Traceback (most recent call last)
Cell In[50], line 23
     20 SMOTENC(categorical_features=[0,2]).fit_resample(Xx.astype({'b':str}), y)                           # OK cat + float + 2str/cat
     21 SMOTENC(categorical_features=[0,2]).fit_resample(Xx.astype({'b':str}).astype({'b':'category'}), y)  # OK cat + float + 2str2cat/cat
---> 23 SMOTENC(categorical_features=[0]).fit_resample(Xx, y)                                             # ERROR cat + float + bool
     24 # SMOTENC(categorical_features=[0,2]).fit_resample(Xx, y)                                           # ERROR cat + float + bool/cat
     25 # SMOTENC(categorical_features=[0,2]).fit_resample(Xx.astype({'b':'category'}), y)                  # ERROR cat + float + 2cat/cat

File ~/miniconda3/lib/python3.10/site-packages/imblearn/base.py:203, in BaseSampler.fit_resample(self, X, y)
    182 """Resample the dataset.
    183 
    184 Parameters
   (...)
    200     The corresponding label of `X_resampled`.
    201 """
    202 self._validate_params()
--> 203 return super().fit_resample(X, y)

File ~/miniconda3/lib/python3.10/site-packages/imblearn/base.py:82, in SamplerMixin.fit_resample(self, X, y)
     80 check_classification_targets(y)
     81 arrays_transformer = ArraysTransformer(X, y)
---> 82 X, y, binarize_y = self._check_X_y(X, y)
     84 self.sampling_strategy_ = check_sampling_strategy(
     85     self.sampling_strategy, y, self._sampling_type
     86 )
     88 output = self._fit_resample(X, y)

File ~/miniconda3/lib/python3.10/site-packages/imblearn/over_sampling/_smote/base.py:540, in SMOTENC._check_X_y(self, X, y)
    536 """Overwrite the checking to let pass some string for categorical
    537 features.
    538 """
    539 y, binarize_y = check_target_type(y, indicate_one_vs_all=True)
--> 540 X, y = self._validate_data(
    541     X, y, reset=True, dtype=None, accept_sparse=["csr", "csc"]
    542 )
    543 return X, y, binarize_y

File ~/miniconda3/lib/python3.10/site-packages/sklearn/base.py:554, in BaseEstimator._validate_data(self, X, y, reset, validate_separately, **check_params)
    552         y = check_array(y, input_name="y", **check_y_params)
    553     else:
--> 554         X, y = check_X_y(X, y, **check_params)
    555     out = X, y
    557 if not no_val_X and check_params.get("ensure_2d", True):

File ~/miniconda3/lib/python3.10/site-packages/sklearn/utils/validation.py:1104, in check_X_y(X, y, accept_sparse, accept_large_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, multi_output, ensure_min_samples, ensure_min_features, y_numeric, estimator)
   1099         estimator_name = _check_estimator_name(estimator)
   1100     raise ValueError(
   1101         f"{estimator_name} requires y to be passed, but the target y is None"
   1102     )
-> 1104 X = check_array(
   1105     X,
   1106     accept_sparse=accept_sparse,
   1107     accept_large_sparse=accept_large_sparse,
   1108     dtype=dtype,
   1109     order=order,
   1110     copy=copy,
   1111     force_all_finite=force_all_finite,
   1112     ensure_2d=ensure_2d,
   1113     allow_nd=allow_nd,
   1114     ensure_min_samples=ensure_min_samples,
   1115     ensure_min_features=ensure_min_features,
   1116     estimator=estimator,
   1117     input_name="X",
   1118 )
   1120 y = _check_y(y, multi_output=multi_output, y_numeric=y_numeric, estimator=estimator)
   1122 check_consistent_length(X, y)

File ~/miniconda3/lib/python3.10/site-packages/sklearn/utils/validation.py:808, in check_array(array, accept_sparse, accept_large_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, ensure_min_samples, ensure_min_features, estimator, input_name)
    803 if pandas_requires_conversion:
    804     # pandas dataframe requires conversion earlier to handle extension dtypes with
    805     # nans
    806     # Use the original dtype for conversion if dtype is None
    807     new_dtype = dtype_orig if dtype is None else dtype
--> 808     array = array.astype(new_dtype)
    809     # Since we converted here, we do not need to convert again later
    810     dtype = None

File ~/miniconda3/lib/python3.10/site-packages/pandas/core/generic.py:6240, in NDFrame.astype(self, dtype, copy, errors)
   6233     results = [
   6234         self.iloc[:, i].astype(dtype, copy=copy)
   6235         for i in range(len(self.columns))
   6236     ]
   6238 else:
   6239     # else, only a single dtype is given
-> 6240     new_data = self._mgr.astype(dtype=dtype, copy=copy, errors=errors)
   6241     return self._constructor(new_data).__finalize__(self, method="astype")
   6243 # GH 33113: handle empty frame or series

File ~/miniconda3/lib/python3.10/site-packages/pandas/core/internals/managers.py:450, in BaseBlockManager.astype(self, dtype, copy, errors)
    449 def astype(self: T, dtype, copy: bool = False, errors: str = "raise") -> T:
--> 450     return self.apply("astype", dtype=dtype, copy=copy, errors=errors)

File ~/miniconda3/lib/python3.10/site-packages/pandas/core/internals/managers.py:352, in BaseBlockManager.apply(self, f, align_keys, ignore_failures, **kwargs)
    350         applied = b.apply(f, **kwargs)
    351     else:
--> 352         applied = getattr(b, f)(**kwargs)
    353 except (TypeError, NotImplementedError):
    354     if not ignore_failures:

File ~/miniconda3/lib/python3.10/site-packages/pandas/core/internals/blocks.py:526, in Block.astype(self, dtype, copy, errors)
    508 """
    509 Coerce to the new dtype.
    510 
   (...)
    522 Block
    523 """
    524 values = self.values
--> 526 new_values = astype_array_safe(values, dtype, copy=copy, errors=errors)
    528 new_values = maybe_coerce_values(new_values)
    529 newb = self.make_block(new_values)

File ~/miniconda3/lib/python3.10/site-packages/pandas/core/dtypes/astype.py:299, in astype_array_safe(values, dtype, copy, errors)
    296     return values.copy()
    298 try:
--> 299     new_values = astype_array(values, dtype, copy=copy)
    300 except (ValueError, TypeError):
    301     # e.g. astype_nansafe can fail on object-dtype of strings
    302     #  trying to convert to float
    303     if errors == "ignore":

File ~/miniconda3/lib/python3.10/site-packages/pandas/core/dtypes/astype.py:227, in astype_array(values, dtype, copy)
    223     return values
    225 if not isinstance(values, np.ndarray):
    226     # i.e. ExtensionArray
--> 227     values = values.astype(dtype, copy=copy)
    229 else:
    230     values = astype_nansafe(values, dtype, copy=copy)

File ~/miniconda3/lib/python3.10/site-packages/pandas/core/arrays/categorical.py:562, in Categorical.astype(self, dtype, copy)
    557     except (
    558         TypeError,  # downstream error msg for CategoricalIndex is misleading
    559         ValueError,
    560     ):
    561         msg = f"Cannot cast {self.categories.dtype} dtype to {dtype}"
--> 562         raise ValueError(msg)
    564     result = take_nd(
    565         new_cats, ensure_platform_int(self._codes), fill_value=fill_value
    566     )
    568 return result

ValueError: Cannot cast object dtype to float64

Versions

Linux-5.4.0-135-generic-x86_64-with-glibc2.31 Python 3.10.8 (main, Nov 24 2022, 14:13:03) [GCC 11.2.0] NumPy 1.22.3 SciPy 1.8.1 Scikit-Learn 1.2.0 Pandas 1.5.2 Imbalanced-Learn 0.10.1

scikit-learn-contrib / imbalanced-learn