(task_type)
| 2351 | |
| 2352 | |
| 2353 | def test_py_data_subgroup_id(task_type): |
| 2354 | train_pool_from_files = Pool(QUERYWISE_TRAIN_FILE, column_description=QUERYWISE_CD_FILE_WITH_SUBGROUP_ID) |
| 2355 | test_pool_from_files = Pool(QUERYWISE_TEST_FILE, column_description=QUERYWISE_CD_FILE_WITH_SUBGROUP_ID) |
| 2356 | model = CatBoostRanker(loss_function='QueryRMSE', iterations=2, thread_count=4, task_type=task_type, gpu_ram_part=TEST_GPU_RAM_PART, devices='0') |
| 2357 | model.fit(train_pool_from_files) |
| 2358 | predictions_from_files = model.predict(test_pool_from_files) |
| 2359 | |
| 2360 | train_df = pd.read_csv(QUERYWISE_TRAIN_FILE, delimiter='\t', header=None) |
| 2361 | train_group_id = train_df.loc[:, 1] |
| 2362 | raw_train_subgroup_id = train_df.loc[:, 4] |
| 2363 | train_target = train_df.loc[:, 2] |
| 2364 | train_data = train_df.drop([0, 1, 2, 3, 4], axis=1).astype(np.float32) |
| 2365 | |
| 2366 | test_df = pd.read_csv(QUERYWISE_TEST_FILE, delimiter='\t', header=None) |
| 2367 | test_data = Pool(test_df.drop([0, 1, 2, 3, 4], axis=1).astype(np.float32)) |
| 2368 | |
| 2369 | for subgroup_id_func in (int, str, lambda id: 'myid_' + str(id)): |
| 2370 | train_subgroup_id = [subgroup_id_func(subgroup_id) for subgroup_id in raw_train_subgroup_id] |
| 2371 | model.fit(train_data, train_target, group_id=train_group_id, subgroup_id=train_subgroup_id) |
| 2372 | predictions_from_py_data = model.predict(test_data) |
| 2373 | assert _check_data(predictions_from_files, predictions_from_py_data) |
| 2374 | |
| 2375 | |
| 2376 | def test_fit_data(task_type): |
nothing calls this directly
no test coverage detected