Shuhei Watanabe comments

Results 34 comments of


                                            Shuhei Watanabe

[refactor] Change names of some variables

https://github.com/automl/Auto-PyTorch/issues/142

torch.Tensor or np.ndarray?

Actually, there are many mypy error regarding wrong typing expressions, so let's fix after the big merge.

[refactor] Separate all_nan_columns detection

Check - test/test_pipeline/components/preprocessing/test_feature_preprocessor.py::TestFeaturePreprocessors::test_pipeline_fit_include[Nystroem-classification_numerical_and_categorical]

[refactor] Separate errors for unsupported columns as a func

Check - test/test_pipeline/components/preprocessing/test_feature_preprocessor.py::TestFeaturePreprocessors::test_pipeline_fit_include[Nystroem-classification_numerical_and_categorical]

Make use of data more by devising subsampling

The information is based on the runs on dev branch. Those passed when I ran with example for the tabular classification. ```python mem_limit, N, D = 3000, 500, 10000 #...

Make use of data more by devising subsampling

The information is based on the runs on common modification branch. ```python mem_limit, N, D = 3000, 500, 10000 # 3GB mem_limit, N, D = 4000, 3000, 10000 # 4GB...

Metric names in additional info are confusing

I did not mean we need to completely rename them because it is painstaking in several ways. 1. Usage in other methods/functions 2. Need to check two names everytime (==>...

[refactor] Separate categorical encoding as a method

- test/test_api/test_base_api.py::test_pipeline_get_budget[3-50-runtime-expected1-classification_categorical_only] - test/test_pipeline/test_tabular_regression.py::TestTabularRegression::test_pipeline_predict[regression_numerical_and_categorical]

[refactor] `_search` function in autoPytorch/api/base_task.py

Suggetions 1. make each function shorter than 40 lines excluding doc-string 2. prohibit the following indent level more than 3 ``` ^ | 40 lines are visible in one screen...

[REFACTOR] Too many variables in `search` and `fit_pipeline`.

> The other exception to the clean code principle might be for front facing API's I agree with this part. However, I do not agree with: > it's okay to...