Benchmark to compare original implemenation vs my vectorized implementation for SMOTE algorithm.

smote_benchmark.py

from time import time

import numpy as np
from scipy import sparse

from imblearn.over_sampling import SMOTE


def benchmark(sampler, X, y):
    imb = np.unique(y, return_counts=True)
    m, n = X.shape
    try:
        d = X.nnz / m / n
        f = X.format
    except AttributeError:
        d = 1
        f = 'arr'
    t0 = time()
    sampler.fit_resample(X, y)
    t = time() - t0
    print(f"{type(sampler).__name__}: X shape: {(m, n)}, y imbalance: {imb}, density: {d}, format: {f}, time: {t}")


sampler = SMOTE(random_state=0)
rng = np.random.RandomState(seed=0)

ms = [1_000, 10_000, 100_000]
ns = [10, 50, 100]
ds = [0.01, 0.1]
ps = [0.01, 0.1]
for m, n in zip(ms, ns):
    for d, p in zip(ds, ps):
        y = rng.choice([0, 1], size=m, p=[p, 1-p])

        X = sparse.random(m, n, density=d, format='coo')
        benchmark(sampler, X, y)

        X = X.tocsr()
        benchmark(sampler, X, y)
        
        X = X.tocsc()
        benchmark(sampler, X, y)
        
        X = X.toarray()
        benchmark(sampler, X, y)

'''
ORIGINAL
--------
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([  7, 993])), density: 0.01, format: coo, time: 0.07027316093444824
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([  7, 993])), density: 0.01, format: csr, time: 0.06946420669555664
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([  7, 993])), density: 0.01, format: csc, time: 0.06998085975646973
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([  7, 993])), density: 1, format: arr, time: 0.005276918411254883
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([ 95, 905])), density: 0.1, format: coo, time: 0.2944509983062744
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([ 95, 905])), density: 0.1, format: csr, time: 0.3965580463409424
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([ 95, 905])), density: 0.1, format: csc, time: 0.30956602096557617
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([ 95, 905])), density: 1, format: arr, time: 0.004606008529663086
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([ 113, 9887])), density: 0.01, format: coo, time: 2.3150041103363037
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([ 113, 9887])), density: 0.01, format: csr, time: 2.228933095932007
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([ 113, 9887])), density: 0.01, format: csc, time: 2.2568271160125732
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([ 113, 9887])), density: 1, format: arr, time: 0.05763578414916992
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([1071, 8929])), density: 0.1, format: coo, time: 3.886691093444824
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([1071, 8929])), density: 0.1, format: csr, time: 3.8926799297332764
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([1071, 8929])), density: 0.1, format: csc, time: 4.220958948135376
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([1071, 8929])), density: 1, format: arr, time: 0.12191629409790039
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([  977, 99023])), density: 0.01, format: coo, time: 32.011396169662476
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([  977, 99023])), density: 0.01, format: csr, time: 32.23091101646423
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([  977, 99023])), density: 0.01, format: csc, time: 33.7115478515625
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([  977, 99023])), density: 1, format: arr, time: 0.7365210056304932
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([10023, 89977])), density: 0.1, format: coo, time: 43.57709002494812
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([10023, 89977])), density: 0.1, format: csr, time: 45.25056982040405
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([10023, 89977])), density: 0.1, format: csc, time: 99.44304585456848
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([10023, 89977])), density: 1, format: arr, time: 16.320679903030396
'''

'''
VECTORIZED
----------
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([  7, 993])), density: 0.01, format: coo, time: 0.0032851696014404297
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([  7, 993])), density: 0.01, format: csr, time: 0.0027887821197509766
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([  7, 993])), density: 0.01, format: csc, time: 0.003389120101928711
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([  7, 993])), density: 1, format: arr, time: 0.0014929771423339844
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([ 95, 905])), density: 0.1, format: coo, time: 0.0031461715698242188
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([ 95, 905])), density: 0.1, format: csr, time: 0.0027971267700195312
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([ 95, 905])), density: 0.1, format: csc, time: 0.0039000511169433594
SMOTE: X shape: (1000, 10), y imbalance: (array([0, 1]), array([ 95, 905])), density: 1, format: arr, time: 0.005051136016845703
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([ 113, 9887])), density: 0.01, format: coo, time: 0.01177215576171875
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([ 113, 9887])), density: 0.01, format: csr, time: 0.008515119552612305
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([ 113, 9887])), density: 0.01, format: csc, time: 0.008799076080322266
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([ 113, 9887])), density: 1, format: arr, time: 0.018984079360961914
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([1071, 8929])), density: 0.1, format: coo, time: 0.05235886573791504
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([1071, 8929])), density: 0.1, format: csr, time: 0.04006767272949219
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([1071, 8929])), density: 0.1, format: csc, time: 0.04768490791320801
SMOTE: X shape: (10000, 50), y imbalance: (array([0, 1]), array([1071, 8929])), density: 1, format: arr, time: 0.09134483337402344
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([  977, 99023])), density: 0.01, format: coo, time: 0.06119227409362793
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([  977, 99023])), density: 0.01, format: csr, time: 0.053843021392822266
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([  977, 99023])), density: 0.01, format: csc, time: 0.08270120620727539
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([  977, 99023])), density: 1, format: arr, time: 0.4651978015899658
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([10023, 89977])), density: 0.1, format: coo, time: 3.597799777984619
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([10023, 89977])), density: 0.1, format: csr, time: 3.4779179096221924
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([10023, 89977])), density: 0.1, format: csc, time: 3.8068392276763916
SMOTE: X shape: (100000, 100), y imbalance: (array([0, 1]), array([10023, 89977])), density: 1, format: arr, time: 16.16456127166748
'''