def test_square_predict(data_random_square):
data = data_random_square
model = SSPOR()
model.fit(data)
sensors = model.get_selected_sensors()
assert data.shape == model.predict(data[:, sensors]).shape
def test_not_fitted(data_vandermonde):
x = data_vandermonde
model = SSPOR()
# Should not be able to call any of these methods before fitting
with pytest.raises(NotFittedError):
model.predict(x)
with pytest.raises(NotFittedError):
model.get_selected_sensors()
with pytest.raises(NotFittedError):
model.get_all_sensors()
with pytest.raises(NotFittedError):
model.set_number_of_sensors(20)
with pytest.raises(NotFittedError):
model.score(x)
with pytest.raises(NotFittedError):
model.reconstruction_error(x)
def test_prefit_basis(data_random):
data = data_random
basis = Identity()
basis.fit(data)
# This data should be ignored during the fit
data_to_ignore = nan * data_random
model = SSPOR(basis=basis)
model.fit(data_to_ignore, prefit_basis=True)
assert not any(isnan(model.get_selected_sensors()))
def test_predict_accuracy(data_vandermonde_testing):
# Polynomials up to degree 10 on [0, 1]
data, x_test = data_vandermonde_testing
model = SSPOR()
model.fit(data, seed=1)
model.set_number_of_sensors(8)
sensors = model.get_selected_sensors()
assert sqrt(mean((x_test - model.predict(x_test[sensors])) ** 2)) <= 1.0e-3
# Should also work for row vectors
x_test = x_test.reshape(1, -1)
assert sqrt(mean((x_test - model.predict(x_test[:, sensors])) ** 2)) <= 1.0e-3
def test_predict(data_random):
data = data_random
n_sensors = 5
model = SSPOR(n_sensors=n_sensors)
model.fit(data)
# Wrong size input for predict
# (should only pass data at sensor locations)
with pytest.raises(ValueError):
model.predict(data)
# Rectangular case
sensors = model.get_selected_sensors()
assert data.shape == model.predict(data[:, sensors]).shape
def test_n_sensors(data_random):
# Check for bad inputs
with pytest.raises(ValueError):
model = SSPOR(n_sensors=0)
with pytest.raises(ValueError):
model = SSPOR(n_sensors=5.4)
with pytest.raises(ValueError):
model = SSPOR(n_sensors="1")
with pytest.raises(ValueError):
model = SSPOR(n_sensors=[1])
n_sensors = 5
x = data_random
model = SSPOR(n_sensors=n_sensors)
model.fit(x)
assert len(model.get_selected_sensors()) == n_sensors
def test_set_number_of_sensors(data_vandermonde):
x = data_vandermonde
max_sensors = x.shape[1]
model = SSPOR()
model.fit(x)
with pytest.raises(ValueError):
model.set_number_of_sensors(max_sensors + 1)
with pytest.raises(ValueError):
model.set_number_of_sensors(0)
with pytest.raises(ValueError):
model.set_number_of_sensors(1.5)
with pytest.raises(ValueError):
model.set_number_of_sensors("3")
model.set_number_of_sensors(15)
assert len(model.get_selected_sensors()) == 15
def test_sensor_selector_properties(data_random):
data = data_random
model = SSPOR().fit(data)
assert all(model.get_all_sensors() == model.all_sensors)
assert all(model.get_selected_sensors() == model.selected_sensors)