This question got to do with SMOTEBoost implementation found here but I believe the issue is relayed to imblearn library.
I tried using the library to re-sample all classes in a multiclass problem. Caught by AttributeError: 'int' object has no attribute 'flatten' error:
How to reproduce the error (in Colab nb):
Clone repo:
!git clone https://github.com/gkapatai/MaatPy.git
cd MaatPy/
from maatpy.classifiers import SMOTEBoost
Dummy data:
X, y = make_classification(n_samples=1000, n_classes=3, n_informative=6, weights=[.1, .15, .75])
xtrain, xtest, ytrain, ytest = train_test_split(X, y, test_size=.2, random_state=123)
Create model:
from maatpy.classifiers import SMOTEBoost
model = SMOTEBoost()
model.fit(xtrain, ytrain)
/usr/local/lib/python3.7/dist-packages/imblearn/over_sampling/_smote.py in _make_samples(self, X, y_dtype, y_type, nn_data, nn_num, n_samples, step_size)
106 random_state = check_random_state(self.random_state)
107 samples_indices = random_state.randint(
--> 108 low=0, high=len(nn_num.flatten()), size=n_samples)
109 steps = step_size * random_state.uniform(size=n_samples)
110 rows = np.floor_divide(samples_indices, nn_num.shape[1])
AttributeError: 'int' object has no attribute 'flatten'
