(task_type)
| 2329 | |
| 2330 | |
| 2331 | def test_py_data_group_id(task_type): |
| 2332 | train_pool_from_files = Pool(QUERYWISE_TRAIN_FILE, column_description=QUERYWISE_CD_FILE_WITH_GROUP_ID) |
| 2333 | test_pool_from_files = Pool(QUERYWISE_TEST_FILE, column_description=QUERYWISE_CD_FILE_WITH_GROUP_ID) |
| 2334 | model = CatBoostRanker(loss_function='QueryRMSE', iterations=2, thread_count=4, task_type=task_type, gpu_ram_part=TEST_GPU_RAM_PART, devices='0') |
| 2335 | model.fit(train_pool_from_files) |
| 2336 | predictions_from_files = model.predict(test_pool_from_files) |
| 2337 | |
| 2338 | train_df = pd.read_csv(QUERYWISE_TRAIN_FILE, delimiter='\t', header=None) |
| 2339 | train_target = train_df.loc[:, 2] |
| 2340 | raw_train_group_id = train_df.loc[:, 1] |
| 2341 | train_data = train_df.drop([0, 1, 2, 3, 4], axis=1).astype(np.float32) |
| 2342 | |
| 2343 | test_df = pd.read_csv(QUERYWISE_TEST_FILE, delimiter='\t', header=None) |
| 2344 | test_data = Pool(test_df.drop([0, 1, 2, 3, 4], axis=1).astype(np.float32)) |
| 2345 | |
| 2346 | for group_id_func in (int, str, lambda id: 'myid_' + str(id)): |
| 2347 | train_group_id = [group_id_func(group_id) for group_id in raw_train_group_id] |
| 2348 | model.fit(train_data, train_target, group_id=train_group_id) |
| 2349 | predictions_from_py_data = model.predict(test_data) |
| 2350 | assert _check_data(predictions_from_files, predictions_from_py_data) |
| 2351 | |
| 2352 | |
| 2353 | def test_py_data_subgroup_id(task_type): |
nothing calls this directly
no test coverage detected