Multi-Input Oversampling

multi-input-oversampling.py

# Sometime we will have multiple input data in different dimensions
# such as wide & deep neural network suggestion model
import numpy as np
from imblearn.over_sampling import RandomOverSampler as ros

Xab = {"a":np.random.random_sample, "b": np.random.random_sample}
Y = [1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
X_i = [[i] for i in range(len(Y))]
X_a = np.array([Xab['a'](1) for y in Y])
X_b = np.array([Xab['b'](2) for y in Y])

# Basically, we use a X_ri in as the resample index to keep the relationship
# in our multiple input data
X_ri, Y_r1 = ros().fit_sample(X_i, Y)

X_a_r = X_a[[i[0] for i in X_ri]]
X_b_r = X_b[[i[0] for i in X_ri]]