def test_naive_bayes_sklearn():
from EvoDAG.naive_bayes import NaiveBayes as NB
from EvoDAG.node import NaiveBayes
try:
from sklearn.naive_bayes import GaussianNB
except ImportError:
return
class Var(object):
def __init__(self, a):
self.hy = a
self.hy_test = None
m = GaussianNB().fit(X, cl)
hy = m._joint_log_likelihood(X)
# hy = m.predict_proba(X) * 2 - 1
vars = [Var(SparseArray.fromlist(x)) for x in X.T]
nb = NB(mask=SparseArray.fromlist([1 for _ in X[:, 0]]),
klass=SparseArray.fromlist(cl),
nclass=3)
naive_bayes = NaiveBayes([x for x in range(4)], naive_bayes=nb)
naive_bayes.eval(vars)
for a, b in zip(naive_bayes.hy, hy.T):
for a1, b1 in zip(a.full_array(), b):
assert_almost_equals(a1, b1, 3)
testRows = tmp
print 'test length = ', len(testRows)
classes = [
'Nominated Best Picture',
'Won Best Picture',
]
clf=GaussianNB()
clf.fit(features[trainRows, :])[:, favoriteCols], labels[trainRows, 0], sample_weight=None)
clf._update_mean_variance(n_past, mu, var, X, sample_weight=None)
clf.partial_fit(features[trainRows, :])[:, favoriteCols], labels[trainRows, 0], classes=classes, sample_weight=None)
clf._partial_fit(features[trainRows, :])[:, favoriteCols], labels[trainRows, 0], classes=classes, _refit=False,sample_weight=None)
clf._joint_log_likelihood(features[trainRows, :])[:, favoriteCols])
print 'accuracy = %f' %(np.mean((y_test-y_pred)==0))
