Out of Core Learning



In [ ]:

    
# write out some toy data
from sklearn.datasets import load_digits
import cPickle

digits = load_digits()

X, y = digits.data, digits.target

for i in range(10):
    cPickle.dump((X[i::10], y[i::10]), open("data/batch_%02d.pickle" % i, "w"), -1)



In [ ]:

    
from sklearn.linear_model import SGDClassifier



In [ ]:

    
sgd = SGDClassifier()

for i in range(9):
    X_batch, y_batch = cPickle.load(open("data/batch_%02d.pickle" % i))
    sgd.partial_fit(X_batch, y_batch, classes=range(10))



In [ ]:

    
X_test, y_test = cPickle.load(open("data/batch_09.pickle"))

sgd.score(X_test, y_test)

Kernel Approximations



In [ ]:

    
from sklearn.kernel_approximation import RBFSampler

sgd = SGDClassifier()
kernel_approximation = RBFSampler(gamma=.001, n_components=400)

for i in range(9):
    X_batch, y_batch = cPickle.load(open("data/batch_%02d.pickle" % i))
    if i == 0:
        kernel_approximation.fit(X_batch)
    X_transformed = kernel_approximation.transform(X_batch)
    sgd.partial_fit(X_transformed, y_batch, classes=range(10))



In [ ]:

    
X_test, y_test = cPickle.load(open("data/batch_09.pickle"))

sgd.score(kernel_approximation.transform(X_test), y_test)

Exercise

Compare the speed and accuracy of LogisticRegression, LinearSVC and SGDClassifier on the bank campaign and adult datasets. Try LogisticRegression(solver='sag').