将分类器保存到scikit-learn中的磁盘

Question

如何将训练有素的朴素贝叶斯分类器保存到磁盘并用于预测数据？我有来自scikit-learn网站的以下示例程序：from sklearn import datasetsiris = datasets.load_iris()from sklearn.naive_bayes import GaussianNBgnb = GaussianNB()y_pred = gnb.fit(iris.data, iris.target).predict(iris.data)print "Number of mislabeled points : %d" % (iris.target != y_pred).sum()

Answer 1

分类器只是可以像其他任何东西一样被腌制和倾倒的对象。继续您的示例：

import cPickle

# save the classifier

with open('my_dumped_classifier.pkl', 'wb') as fid:

    cPickle.dump(gnb, fid)    

# load it again

with open('my_dumped_classifier.pkl', 'rb') as fid:

    gnb_loaded = cPickle.load(fid)

Answer 2

您还可以使用joblib.dump和joblib.load，它们在处理数字数组方面比默认的python pickler效率更高。

Joblib包含在scikit-learn中：

>>> from sklearn.externals import joblib

>>> from sklearn.datasets import load_digits

>>> from sklearn.linear_model import SGDClassifier

>>> digits = load_digits()

>>> clf = SGDClassifier().fit(digits.data, digits.target)

>>> clf.score(digits.data, digits.target)  # evaluate training error

0.9526989426822482

>>> filename = '/tmp/digits_classifier.joblib.pkl'

>>> _ = joblib.dump(clf, filename, compress=9)

>>> clf2 = joblib.load(filename)

>>> clf2

SGDClassifier(alpha=0.0001, class_weight=None, epsilon=0.1, eta0=0.0,

       fit_intercept=True, learning_rate='optimal', loss='hinge', n_iter=5,

       n_jobs=1, penalty='l2', power_t=0.5, rho=0.85, seed=0,

       shuffle=False, verbose=0, warm_start=False)

>>> clf2.score(digits.data, digits.target)

0.9526989426822482