Scribd Logo
Upload

Welcome to Scribd!

  • 3 views

    Experiment 14ml

    Uploaded by

    sailuare

    Copyright:

    © All Rights Reserved
    Download as TXT, PDF, TXT or read online from Scribd

    Experiment 14ml

    Uploaded by

    sailuare
    3 views2 pages

    Original Title

    experiment 14ml

    Copyright

    © © All Rights Reserved

    Available Formats

    TXT, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Copyright:

    © All Rights Reserved
    Download as TXT, PDF, TXT or read online from Scribd
    Download as txt, pdf, or txt
    3 views2 pages

    Experiment 14ml

    Uploaded by

    sailuare

    Copyright:

    © All Rights Reserved
    Download as TXT, PDF, TXT or read online from Scribd
    Download as txt, pdf, or txt
    You are on page 1of 2

    write a program to implement support vector machine and principle compound analysis

    import numpy as np
    from sklearn import datasets, svm
    from sklearn.model_selection import train_test_split
    import matplotlib.pyplot as plt

    def main():
    # Import iris data
    iris = datasets.load_iris()
    x = iris.data[:, :2] # take the first two features.
    y = iris.target # label

    # Splitting data into train, validation and test set


    # 50% training set, 20% validation set, 30% test set
    x_train, x_tmp, y_train, y_tmp = train_test_split(x, y, test_size=0.5,
    random_state=42)
    x_validate, x_test, y_validate, y_test = train_test_split(x_tmp, y_tmp,
    test_size=0.4, random_state=42)

    c = [10**(-3), 10**(- 2), 10**(- 1), 10**0, 10**1, 10**2, 10**3]


    best_values = [-1, -1] # respectively best success rate and best C
    k = 1

    plt.figure('SVC with linear kernel')


    plt.figure(figsize=(30, 20))
    for i in c:
    clf = svm.SVC(kernel='linear', C=i)
    clf.fit(x_train, y_train)

    # create a mesh to plot in


    x_min, x_max = x_train[:, 0].min() - 1, x_train[:, 0].max() + 1
    y_min, y_max = x_train[:, 1].min() - 1, x_train[:, 1].max() + 1
    h = (x_max / x_min) / 100
    xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max,
    h))

    z = clf.predict(np.c_[xx.ravel(), yy.ravel()])
    z = z.reshape(xx.shape)

    plt.subplot(3, 3, k)
    title = 'C=' + str(i)
    plt.title(title, fontsize=30)
    plt.contourf(xx, yy, z, cmap=plt.get('Paired'), alpha=0.8)
    plt.scatter(x_train[:, 0], x_train[:, 1], c=y_train,
    cmap=plt.get('Paired'))
    plt.xlim(xx.min(), xx.max())

    print('With C=' + str(i))


    accuracy = clf.score(x_validate, y_validate)
    print('[-] Accuracy of ' + str(accuracy*100) + '%')
    """
    full_prediction = clf.predict(x_validate)
    print('[-] Number of mislabeled points out of a total %d points: %d'
    % (len(y_validate), (y_validate != full_prediction).sum()))
    accuracy = (100 * (len(y_validate) - (y_validate !=
    full_prediction).sum())) / len(y_validate)
    print('[-] With an accuracy of %.3f%s'
    % (accuracy, '%'))
    """

    if accuracy > best_values[0]:


    best_values = accuracy, i

    k += 1

    plt.show()
    plt.close()

    print('\nBest values of C=' + str(best_values[1]))

    # With the best C evaluating test set


    print('Evaluating test set')
    clf = svm.SVC(kernel='linear', C=best_values[1])
    clf.fit(x_train, y_train)

    x_min, x_max = x_train[:, 0].min() - 1, x_train[:, 0].max() + 1


    y_min, y_max = x_train[:, 1].min() - 1, x_train[:, 1].max() + 1
    h = (x_max / x_min) / 100
    xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))

    z = clf.predict(np.c_[xx.ravel(), yy.ravel()])
    z = z.reshape(xx.shape)

    title = 'C=' + str(best_values[1])


    plt.title(title, fontsize=30)
    plt.contourf(xx, yy, z, cmap=plt.get('Paired'), alpha=0.8)
    plt.scatter(x_train[:, 0], x_train[:, 1], c=y_train, cmap=plt.get('Paired'))
    plt.xlim(xx.min(), xx.max())

    accuracy = clf.score(x_test, y_test)


    print('[-] Accuracy of ' + str(accuracy * 100) + '%')

    """""
    full_prediction = clf.predict(x_test)
    print('[-] Number of mislabeled points out of a total %d points: %d'
    % (len(y_test), (y_test != full_prediction).sum()))
    print('[-] With an accuracy of %.3f%s'
    % ((100 * (len(y_test) - (y_test != full_prediction).sum())) /
    len(y_test), '%'))
    """

    plt.show()
    plt.close()

    if __name__ == "__main__":
    main()

    You might also like

    pFad - Phonifier reborn

    Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

    Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


    Alternative Proxies:

    Alternative Proxy

    pFad Proxy

    pFad v3 Proxy

    pFad v4 Proxy