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