features.SinX1(),
features.SinX2(),
features.SinX3(),
features.SinX4(),
# features.X1Cube(), features.X2Cube(),
# features.X3Cube(), features.X4Cube(),
# features.TanX1(), features.TanX2(),
# features.TanX3(), features.TanX4(),
# features.X1OverX2(), features.X1OverX3(),
# features.X1OverX4(), features.X2OverX3(),
# features.X2OverX4(), features.X3OverX4(),
features.CrossTermX1X2(),
features.CrossTermX1X3(),
features.CrossTermX1X4(),
features.CrossTermX2X3(),
features.CrossTermX2X4(),
features.CrossTermX3X4(),
features.Identity()
])
lm[part].fit(training_input[part], training_output[part])
# Validation
mse_part[part] = lm[part].validate(validation_input[part],
validation_output[part])
print('MSE for part {}: {}'.format(part, mse_part[part]))
print(' ')
print('feature weights for part {} \n{}'.format(part, lm[part].beta))
mse = np.sum(mse_part) / n_parts
print('MSE combined: {}'.format(mse))
