.. _sphx_glr_auto_examples_balance_ex_oversample.py:


=============================
Oversampling minority samples
=============================


This example creates an imbalanced classification dataset, and
oversamples the minority class to balance the class ratios.

.. raw:: html

   <br/>





.. rst-class:: sphx-glr-script-out

 Out::

    Num zero class (pre-balance): 29
    Num one class (pre-balance): 471

    Num zero class (post-balance): 94
    Num one class (post-balance): 471
    Num samples (post-balance): 565
                0  ...        19
    87  -2.153390  ... -0.240686
    367 -0.769973  ... -0.175931
    485 -0.946491  ... -0.120715
    290 -0.839210  ... -0.958555
    72  -1.225000  ...  0.078820

    [5 rows x 20 columns]




|


.. code-block:: python

    print(__doc__)

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

    from sklearn.datasets import make_classification
    from skoot.balance import over_sample_balance
    import pandas as pd

    # #############################################################################
    # Create an imbalanced dataset
    X, y = make_classification(n_samples=500, n_classes=2, weights=[0.05, 0.95],
                               random_state=42)

    # get counts:
    zero_mask = y == 0
    print("Num zero class (pre-balance): %i" % zero_mask.sum())
    print("Num one class (pre-balance): %i\n" % (~zero_mask).sum())

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

    # get the new counts
    new_mask = y_balance == 0
    print("Num zero class (post-balance): %i" % new_mask.sum())
    print("Num one class (post-balance): %i" % (~new_mask).sum())
    print("Num samples (post-balance): %i" % X_balance.shape[0])

    # #############################################################################
    # This also works for pandas DataFrames

    X_balance_df, _ = over_sample_balance(pd.DataFrame.from_records(X),
                                          y, balance_ratio=0.2,
                                          random_state=42)

    print(X_balance_df.head())

**Total running time of the script:** ( 0 minutes  0.031 seconds)



.. only :: html

 .. container:: sphx-glr-footer


  .. container:: sphx-glr-download

     :download:`Download Python source code: ex_oversample.py <ex_oversample.py>`



  .. container:: sphx-glr-download

     :download:`Download Jupyter notebook: ex_oversample.ipynb <ex_oversample.ipynb>`


.. only:: html

 .. rst-class:: sphx-glr-signature

    `Gallery generated by Sphinx-Gallery <https://sphinx-gallery.readthedocs.io>`_