from sklearn import datasets, svm, metrics, cross_validation
from matplotlib import pyplot as pl
import numpy as np
import pickle
%matplotlib inline
%run util.py

digitos = datasets.load_digits()

digitos.data.shape

digitos.data[5]

show_digits_plots(digitos, 0, 9)

show_digits(digitos, 0, 9)

X_train, X_test, y_train, y_test = cross_validation.train_test_split(digitos.data,
                                                                     digitos.target,
                                                                     test_size=.4)

model = svm.SVC(gamma=0.001, C=100.0)
model.fit(X_train, y_train)

pickle.dump(model, open('model', 'w'))

y_predito = model.predict(X_test)

print metrics.classification_report(y_test, y_predito)

cm = metrics.confusion_matrix(y_test, y_predito)

cms = np.zeros((11, 11))
cms[0, :] = np.arange(-1, 10, 1)
cms[:, 0] = np.arange(-1, 10, 1).reshape((11,))
cms[1:, 1:] = cm
print cms

show_confusion_matrix(cm)