def test3():
data = [Entry(id = None, correct = -1, features = np.array([1.0, 1.0])),
Entry(id = None, correct = -1, features = np.array([-1.0, 1.0])),
Entry(id = None, correct = -1, features = np.array([1.0, -1.0])),
Entry(id = None, correct = -1, features = np.array([-1.0, -1.0])),
Entry(id = None, correct = 1, features = np.array([5.0, 5.0])),
Entry(id = None, correct = 1, features = np.array([-5.0, 5.0])),
Entry(id = None, correct = 1, features = np.array([5.0, -5.0])),
Entry(id = None, correct = 1, features = np.array([-5.0, -5.0]))]
classifier = train_svm(data, 100.0, lambda x, y: kernels.poly2(x, y, 0.0))
test_ans = test_svm(data, classifier)
results = calculate_results(data, test_ans)
print(results)
def test3():
data = [
Entry(id=None, correct=-1, features=np.array([1.0, 1.0])),
Entry(id=None, correct=-1, features=np.array([-1.0, 1.0])),
Entry(id=None, correct=-1, features=np.array([1.0, -1.0])),
Entry(id=None, correct=-1, features=np.array([-1.0, -1.0])),
Entry(id=None, correct=1, features=np.array([5.0, 5.0])),
Entry(id=None, correct=1, features=np.array([-5.0, 5.0])),
Entry(id=None, correct=1, features=np.array([5.0, -5.0])),
Entry(id=None, correct=1, features=np.array([-5.0, -5.0]))
]
classifier = train_svm(data, 100.0, lambda x, y: kernels.poly2(x, y, 0.0))
test_ans = test_svm(data, classifier)
results = calculate_results(data, test_ans)
print(results)
results = calculate_results(test_set, test_ans)
err_rate = error_rate(results)
prec = precision(results)
rec = recall(results)
f1 = f1score(results)
if verbose:
print("C = {}, test set error rate = {}".format(bestC, err_rate))
return (classifier, err_rate, prec, rec, f1)
random.seed(6346) # uncomment to make the program deterministic
ks = [(kernels.identity, "Identity kernel"),
(lambda x, y: kernels.poly2(x, y, 0.0), "Homogeneous polynomial kernel")(
lambda x, y: kernels.gaussian(x, y, -0.00001),
"Gaussian kernel, gamma = -0.00001")]
for phi, desc in ks:
cnt = 2
serr = 0.0
sprec = 0.0
srec = 0.0
sf1 = 0.0
print("--------------------------------")
print("Running {} iterations using {}".format(cnt, desc))
for i in range(cnt):
stderr.write("Running... {}/{}\n".format(i, cnt))
_, err_rate, prec, rec, f1 = run_all(data, phi, verbose=True)
err_rate = error_rate(results)
prec = precision(results)
rec = recall(results)
f1 = f1score(results)
if verbose:
print("C = {}, test set error rate = {}".format(bestC, err_rate))
return (classifier, err_rate, prec, rec, f1)
random.seed(6346) # uncomment to make the program deterministic
ks = [
(kernels.identity, "Identity kernel"),
(lambda x, y: kernels.poly2(x, y, 0.0), "Homogeneous polynomial kernel")(
lambda x, y: kernels.gaussian(x, y, -0.00001), "Gaussian kernel, gamma = -0.00001"
),
]
for phi, desc in ks:
cnt = 2
serr = 0.0
sprec = 0.0
srec = 0.0
sf1 = 0.0
print("--------------------------------")
print("Running {} iterations using {}".format(cnt, desc))
for i in range(cnt):
stderr.write("Running... {}/{}\n".format(i, cnt))
