SMOTE class balancer

This example creates an imbalanced classification dataset, and applies the Synthetic Minority Observation TEchnique (SMOTE) class balancing technique in order to balance it by supplementing synthetic minority class observations.

print(__doc__)

# Author: Taylor Smith <taylor.smith@alkaline-ml.com>

from matplotlib import pyplot as plt
from sklearn.datasets import make_blobs
from sklearn.utils.validation import check_random_state
from skoot.balance import smote_balance

plt.figure(figsize=(6, 4))

# #############################################################################
# make blobs and grossly undersample one of them
random_state = check_random_state(42)
n_samples = 1000
X, y = make_blobs(n_samples=n_samples, centers=2,
                  random_state=random_state,
                  n_features=5, cluster_std=4.5)

minority_mask = y == 0
sample_mask = ((~minority_mask) |
               (minority_mask &
                (random_state.rand(minority_mask.shape[0]) < 0.05)))

X, y = X[sample_mask, :], y[sample_mask]

plt.subplot(121)
plt.scatter(X[:, 0], X[:, 1], c=y)
plt.title("Initial imbalanced dataset")

# #############################################################################
# Balance the dataset
X_balance, y_balance = smote_balance(X, y, balance_ratio=0.2, random_state=42)

plt.subplot(122)
plt.scatter(X_balance[:, 0], X_balance[:, 1],
            c=y_balance)
plt.title("Balanced dataset")
plt.show()

Total running time of the script: ( 0 minutes 0.022 seconds)