Boostings implementations eval_set fix #1358
Conversation
|
All PEP8 errors has been fixed, thanks ❤️ Comment last updated at Wed, 22 Jan 2025 21:10:21 |
|
/fix-pep8 |
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #1358 +/- ##
==========================================
- Coverage 80.16% 80.13% -0.03%
==========================================
Files 146 146
Lines 10492 10512 +20
==========================================
+ Hits 8411 8424 +13
- Misses 2081 2088 +7 ☔ View full report in Codecov by Sentry. |
|
Hello @dmitryglhf! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found: There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻 Comment last updated at 2025-01-22 18:09:27 UTC |
| is_classification_task = input_data.task.task_type == TaskTypesEnum.classification | ||
| is_regression_task = input_data.task.task_type == TaskTypesEnum.regression |
There was a problem hiding this comment.
А task_type == ts_forecasting до сюда не может дойти? В любом случае, если случай с обоими false тут некорректен, то стоит его обработать.
There was a problem hiding this comment.
Исправлено: добавлена проверка на уровень выше в модуле boostings.py
| del params["early_stopping_rounds"] | ||
|
|
||
| # Create model with updated params | ||
| if input_data.task.task_type == TaskTypesEnum.classification: |
There was a problem hiding this comment.
Выше уже была такая проверка, можно переиспользовать переменную.
| params = deepcopy(self.model_params) | ||
| if params.get('early_stopping_rounds'): | ||
| del params["early_stopping_rounds"] |
There was a problem hiding this comment.
Возможно лучше вот так менять параметр, через self.params.update?
Это фиксирует то что он поменялся внутри модели.
There was a problem hiding this comment.
Исправлено: исправил на self.params.update.
| train_input, eval_input = train_test_data_setup(input_data) | ||
|
|
||
| # Conditions for training with eval_set | ||
| is_classification_task = input_data.task.task_type == TaskTypesEnum.classification | ||
| is_regression_task = input_data.task.task_type == TaskTypesEnum.regression | ||
| all_classes_present_in_eval = ( | ||
| np.unique(np.array(train_input.target)) in np.unique(np.array(eval_input.target)) | ||
| ) | ||
| is_using_eval_set = bool(self.params.get('use_eval_set')) |
There was a problem hiding this comment.
Если код идентичен тому что выше - можно вынести в функцию.
There was a problem hiding this comment.
Исправлено: вынес в отдельную функцию.
This is a 🔨 code refactoring.
Summary
Resolved issue with
test_preprocessing_through_api/test_categorical_target_processed_correctly. In cases where we tried to use LightGBM, XGBoost or CatBoost an error appeared when usingeval_setdue to the small dataset.