[BUG]sampling_strategy working incorrectly with random oversampling

[ZhiziWen]

[BUG]sampling_strategy working incorrectly with random oversampling

[glemaitre]

We will need a minimal reproducer because I cannot reproduce with a simple example:

from collections import Counter
import numpy as np

n_majority, n_minority = 671, 96
X = np.random.randn(n_majority + n_minority, 10)
y = np.array([0] * n_majority + [1] * n_minority)
Counter(y)

Counter({0: 671, 1: 96})

from imblearn.over_sampling import RandomOverSampler

strategy =  0.25
sampler = RandomOverSampler(sampling_strategy=strategy)
X_res, y_res = sampler.fit_resample(X, y)
Counter(y_res)

Counter({0: 671, 1: 167})

[glemaitre]

Closing since we did not get any reproducer.