def compute_SVM_results(i_train, i_test, n_components=5):
classifiers = []
predictions = []
Xtests = []
ytests = []
Xtrains = []
ytrains = []
for i in range(len(attributes)):
Xtrain = X[i][i_train]
Xtest = X[i][i_test]
ytrain = y[i][i_train]
ytest = y[i][i_test]
clf = GMMBayes(n_components,
min_covar=1E-5,
covariance_type='full',
random_state=0)
clf.fit(Xtrain, ytrain)
y_pred = clf.predict(Xtest)
classifiers.append(clf)
predictions.append(y_pred)
return classifiers, predictions
def gmm_bayes_analysis(X_train, X_test, y_train, y_test):
clf = GMMBayes()
t1 = time.time()
clf.fit(X_train, y_train)
t2 = time.time()
t_train = t2 - t1
t1 = time.time()
score = clf.score(X_test, y_test)
t2 = time.time()
t_test = t2 - t1
# Generate graphs/data for analysis
tpr, fpr, roc_auc = roc_calc(GMMBayes(), X_train, X_test, y_train, y_test)
return tpr, fpr, roc_auc, t_train, t_test, score
def test_too_many_components_warning():
X = np.random.normal(0, 1, size=(3, 2))
y = np.zeros(3)
ncm = 5
clf = GMMBayes(ncm)
with pytest.warns(UserWarning,
match="Expected n_samples >= "
"n_components but got "):
clf.fit(X, y)
def test_incompatible_shapes_exception():
X = np.random.normal(0, 1, size=(100, 2))
y = np.zeros(99)
ncm = 1
clf = GMMBayes(ncm)
with pytest.raises(Exception) as e:
assert clf.fit(X, y)
assert str(e.value) == "X and y have incompatible shapes"
def test_incompatible_number_of_components_exception():
X = np.random.normal(0, 1, size=(100, 2))
y = np.zeros(100)
ncm = [1, 2, 3]
clf = GMMBayes(ncm)
with pytest.raises(Exception) as e:
assert clf.fit(X, y)
assert str(e.value) == ("n_components must be compatible with "
"the number of classes")
def test_gmm1d():
x1 = np.random.normal(0, 1, size=100)
x2 = np.random.normal(10, 1, size=100)
X = np.concatenate((x1, x2)).reshape((200, 1))
y = np.zeros(200)
y[100:] = 1
ncm = 1
clf = GMMBayes(ncm)
clf.fit(X, y)
predicted = clf.predict(X)
assert_allclose(y, predicted)
def test_gmm2d():
x1 = np.random.normal(0, 1, size=(100, 2))
x2 = np.random.normal(10, 1, size=(100, 2))
X = np.vstack((x1, x2))
y = np.zeros(200)
y[100:] = 1
for ncm in (1, 2, 3):
clf = GMMBayes(ncm)
clf.fit(X, y)
predicted = clf.predict(X)
assert_allclose(y, predicted)
def compute_GMMbayes(Ncolors, Ncomp):
classifiers = []
predictions = []
for ncm in Ncomp:
classifiers.append([])
predictions.append([])
for nc in Ncolors:
clf = GMMBayes(ncm, min_covar=1E-5, covariance_type='full')
clf.fit(X_train[:, :nc], y_train)
y_pred = clf.predict(X_test[:, :nc])
classifiers[-1].append(clf)
predictions[-1].append(y_pred)
return classifiers, predictions
#posterior[m] = knc.predict_proba(X_test)
print "Error-Correcting Output Code: ", np.mean(
accuracy) / 0.72, np.std(accuracy) / 0.72
print k
for i in range(0, 6):
for j in range(0, 6):
print '{:5.2f} '.format(box[i, j] / 100.0),
print
#end GNB
box = np.zeros([6, 6])
accuracy = np.zeros(100)
for m in range(0, 100):
gmm = GMMBayes(n_components=3)
y_pred = gmm.fit(X_train, y_train).predict(X_test)
for i in range(0, len(y_pred)):
if y_pred[i] == y_test[i]:
n = n + 1
accuracy[m] = accuracy[m] + 1
box[y_test[i] - 1,
y_pred[i] - 1] = box[y_test[i] - 1, y_pred[i] - 1] + 1
print "Gaussian Mixture Models, n_components=3: ", np.mean(
accuracy) / 0.72, np.std(accuracy) / 0.72
for i in range(0, 6):
for j in range(0, 6):
print '{:5.2f} '.format(box[i, j] / 100.0),
print
box = np.zeros([6, 6])
X_test = sample2
y_test = labels[272:, i]
else:
X_train = training
y_train = labels[:172, i]
X_test = sampletest
y_test = labels[172:, i]
if i >= 2: j = 3
else: j = 6
accuracy = np.zeros(72)
posterior = np.empty([10000, 72, 6])
box = np.zeros([6, 6])
for m in range(0, 100):
gmm = GMMBayes(n_components=6)
gmm.fit(X_train, y_train)
y_pred = gmm.predict(X_test)
n = 0
for i in range(0, len(y_pred)):
if y_pred[i] == y_test[i]:
#print i, y_pred[i], y_test[i]
n = n + 1
accuracy[i] = accuracy[i] + 1
box[y_test[i] - 1,
y_pred[i] - 1] = box[y_test[i] - 1, y_pred[i] - 1] + 1
posterior[m] = gmm.predict_proba(X_test)
print 30, 20, sum(accuracy[0:8]) / 8.0, sum(accuracy[8:18]) / 10.0, sum(
accuracy[18:30]) / 12.0, sum(accuracy[30:43]) / 13.0, sum(
accuracy[43:56]) / 13.0, sum(
