def test_1():
np.random.seed(0)
X = np.random.randn(200, 5)
query_id = np.ones(len(X))
w = np.random.randn(5)
y = np.dot(X, w)
coef, _ = sgd_train(X, y, query_id, 1.0, max_iter=100)
prediction = sgd_predict(X, coef)
tau, _ = stats.kendalltau(y, prediction)
assert np.abs(1 - tau) > 1e-3
def demo():
# Basic demo test
X_train, y_train = load_svmlight_file("demo.train", query_id=False)
X_test, y_test = load_svmlight_file("demo.train", query_id=False)
coef, _ = sgd_train(
X_train, y_train, np.ones(y_test.shape), alpha=0.1, n_features=150000, model="rank", max_iter=100000
)
preds = sgd_predict(X_test, coef, blocks=None)
preds = np.sign(preds)
assert accuracy_score(y_test, preds) > 0.98
assert precision_score(y_test, preds) > 0.98
assert recall_score(y_test, preds) > 0.98
assert auc_score(y_test, preds) > 0.98
