import numpy as np
from rvm import RVMClassification
from sklearn.cross_validation import train_test_split
from sklearn.metrics import accuracy_score
data = np.loadtxt("data_sets/german/german.data-numeric.txt")
for el in data:
if el[-1] == 2:
el[-1] = 0
indexes = [i for i in range(24)]
data_x = data[:, indexes]
data_t = data[:, [24]]
data_t = data_t.reshape((len(data_t), ))
train_x, test_x, train_y, test_y = train_test_split(data_x,
data_t,
test_size=0.3)
cl = RVMClassification(kernel='rbf')
cl.fit(train_x, train_y)
print(accuracy_score(cl.predict(test_x), test_y))
train_x = np.matrix(train_x)
train_y = np.array(train_y)
cl = RVMClassification(kernel="rbf", gamma=1)
cl.fit(train_x, train_y)
valid_x = np.matrix(np.random.uniform(0, 1, size=(50, 2)))
valid_x[:, 1] *= 1.5
valid_x[:, 0] *= np.pi
valid_y = np.sign(np.sin(valid_x[:, 0]) - valid_x[:, 1])
valid_y[valid_y == -1] = 0
predicted_y = cl.predict(valid_x)
plt.plot(train_x[:, 0].A[:n_ex // 2], train_x[:, 1].A[:n_ex // 2], 'rx')
plt.plot(train_x[:, 0].A[n_ex // 2:], train_x[:, 1].A[n_ex // 2:], 'gx')
green_x = []
red_x = []
err = 0
for i in range(len(predicted_y)):
if predicted_y[i] == 1:
red_x.append([valid_x.A[i, 0], valid_x.A[i, 1]])
if valid_y[i] == -1:
err += 1
else:
import numpy as np
from rvm import RVMClassification
from sklearn.cross_validation import train_test_split
from sklearn.metrics import accuracy_score
data = np.loadtxt("data_sets/german/german.data-numeric.txt")
for el in data:
if el[-1] == 2:
el[-1] = 0
indexes = [i for i in range(24)]
data_x = data[:, indexes]
data_t = data[:, [24]]
data_t = data_t.reshape((len(data_t),))
train_x, test_x, train_y, test_y = train_test_split(data_x, data_t, test_size=0.3)
cl = RVMClassification(kernel='rbf')
cl.fit(train_x, train_y)
print(accuracy_score(cl.predict(test_x), test_y))
train_x = np.matrix(train_x)
train_y = np.array(train_y)
cl = RVMClassification(kernel="rbf", gamma=1)
cl.fit(train_x, train_y)
valid_x = np.matrix(np.random.uniform(0, 1, size=(50, 2)))
valid_x[:, 1] *= 1.5
valid_x[:, 0] *= np.pi
valid_y = np.sign(np.sin(valid_x[:, 0]) - valid_x[:, 1])
valid_y[valid_y == -1] = 0
predicted_y = cl.predict(valid_x)
plt.plot(train_x[:, 0].A[:n_ex // 2], train_x[:, 1].A[:n_ex // 2], 'rx')
plt.plot(train_x[:, 0].A[n_ex // 2:], train_x[:, 1].A[n_ex // 2:], 'gx')
green_x = []
red_x = []
err = 0
for i in range(len(predicted_y)):
if predicted_y[i] == 1:
red_x.append([valid_x.A[i, 0], valid_x.A[i, 1]])
if valid_y[i] == -1:
err += 1
else:
