Skip to content

Instantly share code, notes, and snippets.

@giuseppebonaccorso

giuseppebonaccorso/passive_aggressive_regression.py

Created October 6, 2017 10:21
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save giuseppebonaccorso/d459e15308b4faeb3a63bbbf8a6c9462 to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save giuseppebonaccorso/d459e15308b4faeb3a63bbbf8a6c9462 to your computer and use it in GitHub Desktop.
Download ZIP

Revisions

  1. giuseppebonaccorso created this gist Oct 6, 2017.
    44 changes: 44 additions & 0 deletions passive_aggressive_regression.py
    View file
    Original file line number Diff line number Diff line change
    @@ -0,0 +1,44 @@
    import matplotlib.pyplot as plt
    import numpy as np

    from sklearn.datasets import make_regression

    # Set random seed (for reproducibility)
    np.random.seed(1000)

    nb_samples = 500
    nb_features = 4

    # Create the dataset
    X, Y = make_regression(n_samples=nb_samples,
    n_features=nb_features)

    # Implement a Passive Aggressive Regression
    C = 0.01
    eps = 0.1
    w = np.zeros((X.shape[1], 1))
    errors = []

    for i in range(X.shape[0]):
    xi = X[i].reshape((X.shape[1], 1))
    yi = np.dot(w.T, xi)

    loss = max(0, np.abs(yi - Y[i]) - eps)

    tau = loss / (np.power(np.linalg.norm(xi, ord=2), 2) + (1 / (2*C)))

    coeff = tau * np.sign(Y[i] - yi)
    errors.append(np.abs(Y[i] - yi)[0, 0])

    w += coeff * xi

    # Show the error plot
    fig, ax = plt.subplots(figsize=(16, 8))

    ax.plot(errors)
    ax.set_xlabel('Time')
    ax.set_ylabel('Error')
    ax.set_title('Passive Aggressive Regression Absolute Error')
    ax.grid()

    plt.show()