repo_name stringlengths 8 38 | pr_number int64 3 47.1k | pr_title stringlengths 8 175 | pr_description stringlengths 2 19.8k ⌀ | author null | date_created stringlengths 25 25 | date_merged stringlengths 25 25 | filepath stringlengths 6 136 | before_content stringlengths 54 884k ⌀ | after_content stringlengths 56 884k | pr_author stringlengths 3 21 | previous_commit stringlengths 40 40 | pr_commit stringlengths 40 40 | comment stringlengths 2 25.4k | comment_author stringlengths 3 29 | __index_level_0__ int64 0 5.1k |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
scikit-learn-contrib/category_encoders | 398 | (WIP) Partial fix for getting feature names out | I think this is a partial fix for this opened issue:
https://github.com/scikit-learn-contrib/category_encoders/issues/395
It remains to check the behaviour of other estimators that are not ONE_TO_ONE.
Please, let me know if you like the work in progress and I will try to continue. | null | 2023-02-23 13:33:41+00:00 | 2023-03-13 11:48:24+00:00 | tests/test_rankhot.py | import pandas as pd
from unittest import TestCase
import tests.helpers as th
import numpy as np
import category_encoders as encoders
np_X = th.create_array(n_rows=100)
np_X_t = th.create_array(n_rows=50, extras=True)
np_y = np.random.randn(np_X.shape[0]) > 0.5
np_y_t = np.random.randn(np_X_t.shape[0]) > 0.5
X = th.cr... | import pandas as pd
from unittest import TestCase
import tests.helpers as th
import numpy as np
import category_encoders as encoders
np_X = th.create_array(n_rows=100)
np_X_t = th.create_array(n_rows=50, extras=True)
np_y = np.random.randn(np_X.shape[0]) > 0.5
np_y_t = np.random.randn(np_X_t.shape[0]) > 0.5
X = th.cre... | JaimeArboleda | 5eb7a2d6359d680bdadd0534bdb983e712a47f9c | 570827e6b48737d0c9aece8aca31edd6da02c1b2 | why no longer check the dtype? the test hasn't changed except for spacing, has it? why relax it? | PaulWestenthanner | 36 |
scikit-learn-contrib/category_encoders | 398 | (WIP) Partial fix for getting feature names out | I think this is a partial fix for this opened issue:
https://github.com/scikit-learn-contrib/category_encoders/issues/395
It remains to check the behaviour of other estimators that are not ONE_TO_ONE.
Please, let me know if you like the work in progress and I will try to continue. | null | 2023-02-23 13:33:41+00:00 | 2023-03-13 11:48:24+00:00 | tests/test_rankhot.py | import pandas as pd
from unittest import TestCase
import tests.helpers as th
import numpy as np
import category_encoders as encoders
np_X = th.create_array(n_rows=100)
np_X_t = th.create_array(n_rows=50, extras=True)
np_y = np.random.randn(np_X.shape[0]) > 0.5
np_y_t = np.random.randn(np_X_t.shape[0]) > 0.5
X = th.cr... | import pandas as pd
from unittest import TestCase
import tests.helpers as th
import numpy as np
import category_encoders as encoders
np_X = th.create_array(n_rows=100)
np_X_t = th.create_array(n_rows=50, extras=True)
np_y = np.random.randn(np_X.shape[0]) > 0.5
np_y_t = np.random.randn(np_X_t.shape[0]) > 0.5
X = th.cre... | JaimeArboleda | 5eb7a2d6359d680bdadd0534bdb983e712a47f9c | 570827e6b48737d0c9aece8aca31edd6da02c1b2 | I checked if all past tests were still running OK, and I found this particular test failed due to some `dtype` issue. What happened was that one column was of `int32` and the other of `int64` or something like that. But the content was the same. So I added this relaxation so that the test passed. Maybe it depends on so... | JaimeArboleda | 37 |
scikit-learn-contrib/category_encoders | 398 | (WIP) Partial fix for getting feature names out | I think this is a partial fix for this opened issue:
https://github.com/scikit-learn-contrib/category_encoders/issues/395
It remains to check the behaviour of other estimators that are not ONE_TO_ONE.
Please, let me know if you like the work in progress and I will try to continue. | null | 2023-02-23 13:33:41+00:00 | 2023-03-13 11:48:24+00:00 | tests/test_rankhot.py | import pandas as pd
from unittest import TestCase
import tests.helpers as th
import numpy as np
import category_encoders as encoders
np_X = th.create_array(n_rows=100)
np_X_t = th.create_array(n_rows=50, extras=True)
np_y = np.random.randn(np_X.shape[0]) > 0.5
np_y_t = np.random.randn(np_X_t.shape[0]) > 0.5
X = th.cr... | import pandas as pd
from unittest import TestCase
import tests.helpers as th
import numpy as np
import category_encoders as encoders
np_X = th.create_array(n_rows=100)
np_X_t = th.create_array(n_rows=50, extras=True)
np_y = np.random.randn(np_X.shape[0]) > 0.5
np_y_t = np.random.randn(np_X_t.shape[0]) > 0.5
X = th.cre... | JaimeArboleda | 5eb7a2d6359d680bdadd0534bdb983e712a47f9c | 570827e6b48737d0c9aece8aca31edd6da02c1b2 | since you haven't really touched this, might just be on your machine, not on the pipeline. But I'm fine with the relaxation | PaulWestenthanner | 38 |
scikit-learn-contrib/category_encoders | 396 | OneHotEncoder: Adding handle_missing='ignore' option | Closes #386
## Proposed Changes
- added **ignore** option to the `handle_missing` parameter of the `OneHotEncoder`. This will encode `NaN` values as 0 in every dummy column. However, compared to the **value** option, no additional "_nan" category is created.
- added a simple test for the new option. | null | 2023-01-23 15:57:00+00:00 | 2023-01-24 14:38:08+00:00 | tests/test_one_hot.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import numpy as np
import tests.helpers as th
import category_encoders as encoders
class TestOneHotEncoderTestCase(TestCase):
def test_one_hot(self):
X = th.create_dataset(n_rows=100)
X_t = th.cr... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import numpy as np
import tests.helpers as th
import category_encoders as encoders
class TestOneHotEncoderTestCase(TestCase):
def test_one_hot(self):
X = th.create_dataset(n_rows=100)
X_t = th.cr... | woodly0 | 49c62c7b782b04f310a7d48c674b2ee6f3987541 | f66949194ee608e532da5ef86c9078c66e40a145 | could you please add two more columns to the train data set:
- a column containing a `None` instead of a `np.nan`
- a column of type `pd.Categorical`
The often turn out to be edge cases and I think it makes sense to have them tested | PaulWestenthanner | 39 |
scikit-learn-contrib/category_encoders | 396 | OneHotEncoder: Adding handle_missing='ignore' option | Closes #386
## Proposed Changes
- added **ignore** option to the `handle_missing` parameter of the `OneHotEncoder`. This will encode `NaN` values as 0 in every dummy column. However, compared to the **value** option, no additional "_nan" category is created.
- added a simple test for the new option. | null | 2023-01-23 15:57:00+00:00 | 2023-01-24 14:38:08+00:00 | tests/test_one_hot.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import numpy as np
import tests.helpers as th
import category_encoders as encoders
class TestOneHotEncoderTestCase(TestCase):
def test_one_hot(self):
X = th.create_dataset(n_rows=100)
X_t = th.cr... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import numpy as np
import tests.helpers as th
import category_encoders as encoders
class TestOneHotEncoderTestCase(TestCase):
def test_one_hot(self):
X = th.create_dataset(n_rows=100)
X_t = th.cr... | woodly0 | 49c62c7b782b04f310a7d48c674b2ee6f3987541 | f66949194ee608e532da5ef86c9078c66e40a145 | and maybe also an assertion that the mapping is as expected (containing the nan) | PaulWestenthanner | 40 |
scikit-learn-contrib/category_encoders | 396 | OneHotEncoder: Adding handle_missing='ignore' option | Closes #386
## Proposed Changes
- added **ignore** option to the `handle_missing` parameter of the `OneHotEncoder`. This will encode `NaN` values as 0 in every dummy column. However, compared to the **value** option, no additional "_nan" category is created.
- added a simple test for the new option. | null | 2023-01-23 15:57:00+00:00 | 2023-01-24 14:38:08+00:00 | tests/test_one_hot.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import numpy as np
import tests.helpers as th
import category_encoders as encoders
class TestOneHotEncoderTestCase(TestCase):
def test_one_hot(self):
X = th.create_dataset(n_rows=100)
X_t = th.cr... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import numpy as np
import tests.helpers as th
import category_encoders as encoders
class TestOneHotEncoderTestCase(TestCase):
def test_one_hot(self):
X = th.create_dataset(n_rows=100)
X_t = th.cr... | woodly0 | 49c62c7b782b04f310a7d48c674b2ee6f3987541 | f66949194ee608e532da5ef86c9078c66e40a145 | Thanks for your feedback. I've extended the existing test and added an additional one. | woodly0 | 41 |
scikit-learn-contrib/category_encoders | 381 | [DOC] Catboost docs reformulation | Connected to #337 and #351 | null | 2022-10-31 04:47:51+00:00 | 2022-11-01 20:33:10+00:00 | category_encoders/cat_boost.py | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | glevv | c5dd2b7ac414eedda555b74207223beb2c557b05 | 0fd5d2836366237689bbb89699232e2f6acb40d6 | there is a typo heere `CatBoosts` | PaulWestenthanner | 42 |
scikit-learn-contrib/category_encoders | 381 | [DOC] Catboost docs reformulation | Connected to #337 and #351 | null | 2022-10-31 04:47:51+00:00 | 2022-11-01 20:33:10+00:00 | category_encoders/cat_boost.py | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | glevv | c5dd2b7ac414eedda555b74207223beb2c557b05 | 0fd5d2836366237689bbb89699232e2f6acb40d6 | Should we add that if you're not sure whether information leaks you can do a random permutation first? | PaulWestenthanner | 43 |
scikit-learn-contrib/category_encoders | 381 | [DOC] Catboost docs reformulation | Connected to #337 and #351 | null | 2022-10-31 04:47:51+00:00 | 2022-11-01 20:33:10+00:00 | category_encoders/cat_boost.py | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | glevv | c5dd2b7ac414eedda555b74207223beb2c557b05 | 0fd5d2836366237689bbb89699232e2f6acb40d6 | I don't like the word `continue` here. The encoder is newly fitted according to target information.
If I do something like
``` python
enc = CatBoostEncoder()
enc.fit_transform(X1, y1)
enc.fit_transform(X2, y2)
```
the previous fit will be ignored. So `continue` might be misleading and `refit` might be a bett... | PaulWestenthanner | 44 |
scikit-learn-contrib/category_encoders | 381 | [DOC] Catboost docs reformulation | Connected to #337 and #351 | null | 2022-10-31 04:47:51+00:00 | 2022-11-01 20:33:10+00:00 | category_encoders/cat_boost.py | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | glevv | c5dd2b7ac414eedda555b74207223beb2c557b05 | 0fd5d2836366237689bbb89699232e2f6acb40d6 | I reworded it, I think it should be clearer | glevv | 45 |
scikit-learn-contrib/category_encoders | 373 | Target encoding heirarchical columnwise | This pull request enhances hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress
## Proposed Changes
Allows a user to submit a hierarchy within a dataframe (i.e. columnwise), not just a mapping dictionary.
Columns must take the names HIER_colA_1, HIER_colA_2, HIER_colA_3, HIER_co... | null | 2022-10-04 08:10:13+00:00 | 2022-10-05 13:32:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | nercisla | 81bb01d99a44624f117cf13bb7ef64ef55ee7f9d | a0d4748d1ecb6b343db079a42133a1d47263fa49 | maybe add the corresponding import statement to make it a self contained example
``` >>> from category_encoders.datasets import load_compass``
| PaulWestenthanner | 46 |
scikit-learn-contrib/category_encoders | 373 | Target encoding heirarchical columnwise | This pull request enhances hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress
## Proposed Changes
Allows a user to submit a hierarchy within a dataframe (i.e. columnwise), not just a mapping dictionary.
Columns must take the names HIER_colA_1, HIER_colA_2, HIER_colA_3, HIER_co... | null | 2022-10-04 08:10:13+00:00 | 2022-10-05 13:32:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | nercisla | 81bb01d99a44624f117cf13bb7ef64ef55ee7f9d | a0d4748d1ecb6b343db079a42133a1d47263fa49 | why do we need this error? the current behaviour is that if you do not specify `cols` all categorical columns will be determined and encoded. So if i rely on this default and not specify columns this will break now right? | PaulWestenthanner | 47 |
scikit-learn-contrib/category_encoders | 373 | Target encoding heirarchical columnwise | This pull request enhances hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress
## Proposed Changes
Allows a user to submit a hierarchy within a dataframe (i.e. columnwise), not just a mapping dictionary.
Columns must take the names HIER_colA_1, HIER_colA_2, HIER_colA_3, HIER_co... | null | 2022-10-04 08:10:13+00:00 | 2022-10-05 13:32:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | nercisla | 81bb01d99a44624f117cf13bb7ef64ef55ee7f9d | a0d4748d1ecb6b343db079a42133a1d47263fa49 | ah ok, in line 138 you have to know at init time what the columns will look like, determining them on the fly during fit-time will not be enough | PaulWestenthanner | 48 |
scikit-learn-contrib/category_encoders | 373 | Target encoding heirarchical columnwise | This pull request enhances hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress
## Proposed Changes
Allows a user to submit a hierarchy within a dataframe (i.e. columnwise), not just a mapping dictionary.
Columns must take the names HIER_colA_1, HIER_colA_2, HIER_colA_3, HIER_co... | null | 2022-10-04 08:10:13+00:00 | 2022-10-05 13:32:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | nercisla | 81bb01d99a44624f117cf13bb7ef64ef55ee7f9d | a0d4748d1ecb6b343db079a42133a1d47263fa49 | I think we should document somewhere that in case of supplying a dataframe as hierarchy that the columns have to be called HIER_col_1, ... HIER_col_N.
Also I'm not sure if 1 or N is the coarsest level (so probably also worth documenting) | PaulWestenthanner | 49 |
scikit-learn-contrib/category_encoders | 373 | Target encoding heirarchical columnwise | This pull request enhances hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress
## Proposed Changes
Allows a user to submit a hierarchy within a dataframe (i.e. columnwise), not just a mapping dictionary.
Columns must take the names HIER_colA_1, HIER_colA_2, HIER_colA_3, HIER_co... | null | 2022-10-04 08:10:13+00:00 | 2022-10-05 13:32:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | nercisla | 81bb01d99a44624f117cf13bb7ef64ef55ee7f9d | a0d4748d1ecb6b343db079a42133a1d47263fa49 | this check is not specific to hierarchy right? if so it should be moved to the base encoder. Strange that we do not have such a check implemented already... | PaulWestenthanner | 50 |
scikit-learn-contrib/category_encoders | 373 | Target encoding heirarchical columnwise | This pull request enhances hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress
## Proposed Changes
Allows a user to submit a hierarchy within a dataframe (i.e. columnwise), not just a mapping dictionary.
Columns must take the names HIER_colA_1, HIER_colA_2, HIER_colA_3, HIER_co... | null | 2022-10-04 08:10:13+00:00 | 2022-10-05 13:32:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | nercisla | 81bb01d99a44624f117cf13bb7ef64ef55ee7f9d | a0d4748d1ecb6b343db079a42133a1d47263fa49 | in the postcode example cols should be just `postcode` right? so specifying `cols=None` would not make any sense anyway since the HIER_postcode_N cols would be encoded as columns on their own instead of being recognized in the hierarchy.
Then this is just fine to throw an error here | PaulWestenthanner | 51 |
scikit-learn-contrib/category_encoders | 373 | Target encoding heirarchical columnwise | This pull request enhances hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress
## Proposed Changes
Allows a user to submit a hierarchy within a dataframe (i.e. columnwise), not just a mapping dictionary.
Columns must take the names HIER_colA_1, HIER_colA_2, HIER_colA_3, HIER_co... | null | 2022-10-04 08:10:13+00:00 | 2022-10-05 13:32:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | nercisla | 81bb01d99a44624f117cf13bb7ef64ef55ee7f9d | a0d4748d1ecb6b343db079a42133a1d47263fa49 | Agreed. We have moved this, hopefully to the right place. | nercisla | 52 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | there seems to be a typo throughout the whole pull request, hierarchy is spelled with e and i switched right? | PaulWestenthanner | 53 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | please also add the parameter to the docstring with a little explanation and type (guess it is dict mapping cols to the hierarchy levels right?) | PaulWestenthanner | 54 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | I don't think copying is needed here | PaulWestenthanner | 55 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | `self.cols` gets declared in the `fit` method of the `BaseEncoder` before calling `_fit`. So it will be already declared at this point. I'm not sure if the additional columns that are created on the fly need to be added here as well. | PaulWestenthanner | 56 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | does this work for a multi-hierarchy as well? e.g. imagine in the animal example you have also some insects and another super-category `mammal` which both `Feline` and `Canine`belong to but not the insects. I think this would be at least dependent on the order of the dictionary (which we don't want since dicts are gene... | PaulWestenthanner | 57 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | better use `col.startswith` as this is more precise | PaulWestenthanner | 58 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | what do you mean by "third dimension"? The dimension in the ordinal encoder are the number of columns that are encoded. So subtracting 1 here should be fine | PaulWestenthanner | 59 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | I would find this more readable if you write `prior` instead of `scalar`. This should be the same right? | PaulWestenthanner | 60 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | this also seems to work for single-level hierarchies only, right? So at some point we'd need to raise an exception (probably when parsing the hierarchy map) | PaulWestenthanner | 61 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | we had a discussion in this project recently where a pandas guy discouraged using `inplace` and actually removed all (or at least most of) `inplace` usages from this repo. I don't remember the exact reason but it made sense to me and we should probably follow suit here as well. So just use
`X = X.drop(...)` | PaulWestenthanner | 62 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | This is basically cleaning the ordinal encoder so that it never new there was hierarchy involved? I like the idea that all the additional columns should be hidden from the user | PaulWestenthanner | 63 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | Yes, you are correct. Though the word heir is spelt with i and e the opposite way around. ;) | nercisla | 64 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | No, it's not needed. It was a remnant of an old solution that still required checking. | nercisla | 65 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | This is also be a remnant of an old solution. | nercisla | 66 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | I have added this now. | nercisla | 67 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | Working on fixing the tests, Joe and I found a problem here, so the method is a little different now. We separate the OrdinalEncoders for the base column(s) and the hierarchy. | nercisla | 68 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | At this stage, we only consider one level of hierarchy, but know how to change it, so can do that soon. Either this release or another. | nercisla | 69 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | Agreed. | nercisla | 70 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | No, I don't believe that is true. The prior is taken over all the data, but the scalar is an estimate of the probability. The paper clearly defines it separately, so feel the language should follow. | nercisla | 71 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | Agreed. | nercisla | 72 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | No problem. Though nicely this line has been dropped due to the cleaned up methodology :) | nercisla | 73 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | As above, this line has been removed in the new method. | nercisla | 74 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | I'm fine with having it in another release. Let's get the first working version merged soon and then iterate from there | PaulWestenthanner | 75 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | The variable called `scalar_heir` is defined separately and should keep its name. My comment war referring to the variable called `scalar` on the same line. This is even set equal to `prior` in line 155. | PaulWestenthanner | 76 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | So, the scalar is normally set to the prior, but for multiple hierarchies, the scalar is a function. This matters more when we will have multi-level hierarchies, in which the "scalar" will definitely move away from the prior. So, in that sense, I think it matters that we don't call it the prior. The prior and scalar... | nercisla | 77 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | this library needs to be added to the setup.py as well.
Do you think its possible to flatten the dictionary without a dependency to another library? I've quickly checked it and it does not seem to be super actively maintained (latest release is over one year ago and it officially does not support python 3.9 onwards) | PaulWestenthanner | 78 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | I don't like a capital D as variable name, something like `flattened_hierarchy` seems more telling | PaulWestenthanner | 79 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | also in the `True` case you want to return the size right? otherwise `hierarchy_check[1]` might be undefined (c.f. line 113) | PaulWestenthanner | 80 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | also the one-line statement seems bulky and I'd find a multiline statement more readable (but this might just be personal taste)
```
if min_tuple_size == max_tuple_size:
return ...
else:
return ....
`` | PaulWestenthanner | 81 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | checking for types should be done using `isinstance(t, tuple)` | PaulWestenthanner | 82 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | an f-string would be more readable here | PaulWestenthanner | 83 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | this is also the reason why the tests are failing at the moment | PaulWestenthanner | 84 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | you're right! thanks for that explanation | PaulWestenthanner | 85 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | I think the flatten-dict function is complete and possibly too extensive to re-write. We could ask the author if we could copy the required function into our utils, but I'd hesitate to re-write it. We tried and found flatten-dict to be better. Will add it into setiup.py for the moment, until we reach concensus. | nercisla | 86 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | Will change this! | nercisla | 87 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | > also in the `True` case you want to return the size right? otherwise `hierarchy_check[1]` might be undefined (c.f. line 113)
This actually does the job despite what it seems at first glance. The if/else come first then the min_size is added to either afterwards. So it still returns the min_size in both cases. | nercisla | 88 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | But I agree it's not easily readable (even for me second time round) so will change it. | nercisla | 89 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | Agreed. | nercisla | 90 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | aaah that's how it works. Thanks for changing anyway | PaulWestenthanner | 91 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | I think I just found a neat way:
pandas apparently implements a function `json_normalize` that can be used to flatten dicts as well. This is very well maintained. For reference it's way number 3 here: https://www.freecodecamp.org/news/how-to-flatten-a-dictionary-in-python-in-4-different-ways/
I just tried it out an... | PaulWestenthanner | 92 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
Supported targets: ... | """Target Encoder"""
import numpy as np
import pandas as pd
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
import warnings
__author__ = 'chappers'
class TargetEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""Target encoding for categorical features.
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | A very nice solution! I have tested it, so will push it to the branch now. Nice work! | nercisla | 93 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | tests/test_target_encoder.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def test_target_encoder(self):
np_X = th.create_array(n_rows=100)
np_X_t = th... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def setUp(self):
self.hierarchical_cat_example = pd.DataFrame(
{
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | this will just overwrite lines 117-125 | PaulWestenthanner | 94 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | tests/test_target_encoder.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def test_target_encoder(self):
np_X = th.create_array(n_rows=100)
np_X_t = th... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def setUp(self):
self.hierarchical_cat_example = pd.DataFrame(
{
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | another nice test would be to check with a "trivial hierarchy", i.e. mapping everything to a unique top level does the same thing as doing target encoding without any hierarchy | PaulWestenthanner | 95 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | tests/test_target_encoder.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def test_target_encoder(self):
np_X = th.create_array(n_rows=100)
np_X_t = th... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def setUp(self):
self.hierarchical_cat_example = pd.DataFrame(
{
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | and maybe another test for multi-level hierarchies where we'd expect an error. | PaulWestenthanner | 96 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | tests/test_target_encoder.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def test_target_encoder(self):
np_X = th.create_array(n_rows=100)
np_X_t = th... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def setUp(self):
self.hierarchical_cat_example = pd.DataFrame(
{
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | Correct. | nercisla | 97 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | tests/test_target_encoder.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def test_target_encoder(self):
np_X = th.create_array(n_rows=100)
np_X_t = th... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def setUp(self):
self.hierarchical_cat_example = pd.DataFrame(
{
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | Yup, agreed, as this currently breaks, I think. | nercisla | 98 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | tests/test_target_encoder.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def test_target_encoder(self):
np_X = th.create_array(n_rows=100)
np_X_t = th... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def setUp(self):
self.hierarchical_cat_example = pd.DataFrame(
{
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | Do you mean a test to check it errors elegantly? | nercisla | 99 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | tests/test_target_encoder.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def test_target_encoder(self):
np_X = th.create_array(n_rows=100)
np_X_t = th... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def setUp(self):
self.hierarchical_cat_example = pd.DataFrame(
{
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | yes, just an `self.assertRaises` statement that checks for the error of mulit-level hierarchies | PaulWestenthanner | 100 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | tests/test_target_encoder.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def test_target_encoder(self):
np_X = th.create_array(n_rows=100)
np_X_t = th... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def setUp(self):
self.hierarchical_cat_example = pd.DataFrame(
{
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | I just saw you've already implemented that. Very nice! | PaulWestenthanner | 101 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | tests/test_target_encoder.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def test_target_encoder(self):
np_X = th.create_array(n_rows=100)
np_X_t = th... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def setUp(self):
self.hierarchical_cat_example = pd.DataFrame(
{
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | Why do you think this breaks? According to the paper this should be fine, right? | PaulWestenthanner | 102 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | tests/test_target_encoder.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def test_target_encoder(self):
np_X = th.create_array(n_rows=100)
np_X_t = th... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def setUp(self):
self.hierarchical_cat_example = pd.DataFrame(
{
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | I just tried it out and it works fine. Here's a complete test. I'd be happy if you include it
```python
def test_trivial_hierarchy(self):
trivial_hierarchical_map = {
'Plant': {
'Plant': ('Rose', 'Daisy', 'Daffodil', 'Bluebell')
}
}
enc_hie... | PaulWestenthanner | 103 |
scikit-learn-contrib/category_encoders | 366 | Target encoding hierarchical | Fixes #136
## Proposed Changes
This pull request implements feature hierarchies in Target Encoders.
Author: @nercisla
Current status: Work in Progress | null | 2022-08-07 07:41:06+00:00 | 2022-09-06 16:18:16+00:00 | tests/test_target_encoder.py | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def test_target_encoder(self):
np_X = th.create_array(n_rows=100)
np_X_t = th... | import pandas as pd
from unittest import TestCase # or `from unittest import ...` if on Python 3.4+
import tests.helpers as th
import numpy as np
import category_encoders as encoders
class TestTargetEncoder(TestCase):
def setUp(self):
self.hierarchical_cat_example = pd.DataFrame(
{
... | PaulWestenthanner | 6a13c14919d56fed8177a173d4b3b82c5ea2fef5 | 30f05c0b584b453332a97389494f95b218495d32 | @nercisla have you seen this comment? | PaulWestenthanner | 104 |
scikit-learn-contrib/category_encoders | 360 | Upgrade versions |
Fixes #359
## Proposed Changes
- drop support for python 3.5 and python 3.6
- drop support for pandas <1.0.5
- introduce f strings
| null | 2022-05-30 16:24:17+00:00 | 2022-05-31 12:57:43+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'chappers'
class TargetEncoder(BaseEstimator, util.TransformerWithTargetMixin):
"""Target encoding for catego... | """Target Encoder"""
import warnings
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'chappers'
class TargetEncoder(BaseEstimator, util.TransformerWithTargetMixin):
"""Target enc... | PaulWestenthanner | 3ae37104c2a5884f701d518fab234a3cb8859dd0 | a18cb64a81310a5e515c7b21255597b4dfb29b86 | Isn't this a typo? Should be `if smoothing == 1.0` right? | freddyaboulton | 105 |
scikit-learn-contrib/category_encoders | 360 | Upgrade versions |
Fixes #359
## Proposed Changes
- drop support for python 3.5 and python 3.6
- drop support for pandas <1.0.5
- introduce f strings
| null | 2022-05-30 16:24:17+00:00 | 2022-05-31 12:57:43+00:00 | category_encoders/target_encoder.py | """Target Encoder"""
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'chappers'
class TargetEncoder(BaseEstimator, util.TransformerWithTargetMixin):
"""Target encoding for catego... | """Target Encoder"""
import warnings
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'chappers'
class TargetEncoder(BaseEstimator, util.TransformerWithTargetMixin):
"""Target enc... | PaulWestenthanner | 3ae37104c2a5884f701d518fab234a3cb8859dd0 | a18cb64a81310a5e515c7b21255597b4dfb29b86 | yes this is a typo. You're right. I noticed it myself and already fixed it in this commit https://github.com/scikit-learn-contrib/category_encoders/commit/aed0fb2ce5a7ed2b11d04e891e523411c5f0e5fa. | PaulWestenthanner | 106 |
scikit-learn-contrib/category_encoders | 334 | Fix pandas future warning for dropping invariants | When I use `drop_invariant=True`, I get a `FutureWarning` because `df.drop` is switching to named kwargs only. This fixes it.
It should also be substantially more performant as the DF is only modified once. | null | 2022-01-18 14:07:13+00:00 | 2022-02-14 07:40:26+00:00 | category_encoders/backward_difference.py | """Backward difference contrast encoding"""
import pandas as pd
from sklearn.base import BaseEstimator, TransformerMixin
from patsy.contrasts import Diff
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class BackwardDifferenc... | """Backward difference contrast encoding"""
import pandas as pd
from sklearn.base import BaseEstimator, TransformerMixin
from patsy.contrasts import Diff
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class BackwardDifferenc... | jona-sassenhagen | 531a271424fcea26df380709ac5700fb2d88c33c | d737e1758b5b206286251cf79cf35f1496154e56 | Why'd you remove `inplace=True` here? | zachmayer | 107 |
scikit-learn-contrib/category_encoders | 334 | Fix pandas future warning for dropping invariants | When I use `drop_invariant=True`, I get a `FutureWarning` because `df.drop` is switching to named kwargs only. This fixes it.
It should also be substantially more performant as the DF is only modified once. | null | 2022-01-18 14:07:13+00:00 | 2022-02-14 07:40:26+00:00 | category_encoders/backward_difference.py | """Backward difference contrast encoding"""
import pandas as pd
from sklearn.base import BaseEstimator, TransformerMixin
from patsy.contrasts import Diff
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class BackwardDifferenc... | """Backward difference contrast encoding"""
import pandas as pd
from sklearn.base import BaseEstimator, TransformerMixin
from patsy.contrasts import Diff
import numpy as np
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class BackwardDifferenc... | jona-sassenhagen | 531a271424fcea26df380709ac5700fb2d88c33c | d737e1758b5b206286251cf79cf35f1496154e56 | I think the pandas dev team is not very happy with `inplace` and there's semi-regular attempts to get rid of it in general. I personally think `inplace` is not the most readable and transparent way to write pandas code and I try to avoid it.
There's an extensive discussion here: https://github.com/pandas-dev/pandas/... | jona-sassenhagen | 108 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/binary.py | """Binary encoding"""
import pandas as pd
from sklearn.base import BaseEstimator, TransformerMixin
import category_encoders as ce
__author__ = 'willmcginnis'
class BinaryEncoder(BaseEstimator, TransformerMixin):
"""Binary encoding for categorical variables, similar to onehot, but stores categories as binary bi... | """Binary encoding"""
from functools import partialmethod
from category_encoders import utils
from category_encoders.basen import BaseNEncoder
__author__ = 'willmcginnis'
class BinaryEncoder(BaseNEncoder):
"""Binary encoding for categorical variables, similar to onehot, but stores categories as binary bitstrings... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | I think it should be fine, but have you checked that cloning and get/set params work? | bmreiniger | 109 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/binary.py | """Binary encoding"""
import pandas as pd
from sklearn.base import BaseEstimator, TransformerMixin
import category_encoders as ce
__author__ = 'willmcginnis'
class BinaryEncoder(BaseEstimator, TransformerMixin):
"""Binary encoding for categorical variables, similar to onehot, but stores categories as binary bi... | """Binary encoding"""
from functools import partialmethod
from category_encoders import utils
from category_encoders.basen import BaseNEncoder
__author__ = 'willmcginnis'
class BinaryEncoder(BaseNEncoder):
"""Binary encoding for categorical variables, similar to onehot, but stores categories as binary bitstrings... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | what do you mean by cloning? I just checked get/set params works. Also tests still work | PaulWestenthanner | 110 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/binary.py | """Binary encoding"""
import pandas as pd
from sklearn.base import BaseEstimator, TransformerMixin
import category_encoders as ce
__author__ = 'willmcginnis'
class BinaryEncoder(BaseEstimator, TransformerMixin):
"""Binary encoding for categorical variables, similar to onehot, but stores categories as binary bi... | """Binary encoding"""
from functools import partialmethod
from category_encoders import utils
from category_encoders.basen import BaseNEncoder
__author__ = 'willmcginnis'
class BinaryEncoder(BaseNEncoder):
"""Binary encoding for categorical variables, similar to onehot, but stores categories as binary bitstrings... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | `sklearn.base.clone` gets used to copy an estimator, keeping hyperparameters but dropping fitted parameters. It uses `get_params` and then instantiates a new class of the same type using the gotten parameters, so I expect this is fine if `get_params` works. | bmreiniger | 111 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/cat_boost.py | """CatBoost coding"""
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(BaseEstimator, util.TransformerWithTargetMixin):
"""CatBoost Encoding for c... | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | ```suggestion
nan_cond = is_nan & unseen_values.isnull().any()
if nan_cond.any():
X.loc[nan_cond, col] = self._mean
``` | jona-sassenhagen | 112 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/cat_boost.py | """CatBoost coding"""
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(BaseEstimator, util.TransformerWithTargetMixin):
"""CatBoost Encoding for c... | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | Isn't this just
```suggestion
return_map = pd.Series(dict(enumerate(categories)))
``` | jona-sassenhagen | 113 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/cat_boost.py | """CatBoost coding"""
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(BaseEstimator, util.TransformerWithTargetMixin):
"""CatBoost Encoding for c... | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | thanks for that suggestion. I've implemented it | PaulWestenthanner | 114 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/cat_boost.py | """CatBoost coding"""
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(BaseEstimator, util.TransformerWithTargetMixin):
"""CatBoost Encoding for c... | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | thanks! | PaulWestenthanner | 115 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/cat_boost.py | """CatBoost coding"""
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(BaseEstimator, util.TransformerWithTargetMixin):
"""CatBoost Encoding for c... | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | You've added a tag `predict_depends_on_y`, but don't seem to use it anywhere; is there a later purpose this serves? Shouldn't it be `transform_depends_on_y` anyway? | bmreiniger | 116 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/cat_boost.py | """CatBoost coding"""
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(BaseEstimator, util.TransformerWithTargetMixin):
"""CatBoost Encoding for c... | """CatBoost coding"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from sklearn.utils.random import check_random_state
__author__ = 'Jan Motl'
class CatBoostEncoder(util.BaseEncoder, util.SupervisedTransformerMixin):
"""CatBoost Encoding for categorical features.
Supported ... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | This was also introduced for "dynamic docstrings" and kept since it introduces useful information (also here you could imagine a test, also testing that the encoders where transform does not depend on y actually give the same result with and without y).
I agree that `transform_depends_on_y` is a more suitable name | PaulWestenthanner | 117 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/count.py | """Count Encoder"""
from __future__ import division
import numpy as np
import pandas as pd
import category_encoders.utils as util
from category_encoders.ordinal import OrdinalEncoder
from copy import copy
from sklearn.base import BaseEstimator, TransformerMixin
__author__ = 'joshua t. dunn'
class CountEncoder(Bas... | """Count Encoder"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from category_encoders.ordinal import OrdinalEncoder
from copy import copy
__author__ = 'joshua t. dunn'
class CountEncoder(util.BaseEncoder, util.UnsupervisedTransformerMixin):
prefit_ordinal = True
encoding_... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | ```suggestion
X[col] = X[col].fillna(value=np.nan)
```
should be much more efficient no? | jona-sassenhagen | 118 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/count.py | """Count Encoder"""
from __future__ import division
import numpy as np
import pandas as pd
import category_encoders.utils as util
from category_encoders.ordinal import OrdinalEncoder
from copy import copy
from sklearn.base import BaseEstimator, TransformerMixin
__author__ = 'joshua t. dunn'
class CountEncoder(Bas... | """Count Encoder"""
import numpy as np
import pandas as pd
import category_encoders.utils as util
from category_encoders.ordinal import OrdinalEncoder
from copy import copy
__author__ = 'joshua t. dunn'
class CountEncoder(util.BaseEncoder, util.UnsupervisedTransformerMixin):
prefit_ordinal = True
encoding_... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | absolutely right! thanks | PaulWestenthanner | 119 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/utils.py | """A collection of shared utilities for all encoders, not intended for external use."""
import pandas as pd
import numpy as np
from scipy.sparse import csr_matrix
__author__ = 'willmcginnis'
def convert_cols_to_list(cols):
if isinstance(cols, pd.Series):
return cols.tolist()
elif isinstance(cols, np... | """A collection of shared utilities for all encoders, not intended for external use."""
from abc import abstractmethod
from enum import Enum, auto
import pandas as pd
import numpy as np
import sklearn.base
from sklearn.base import BaseEstimator, TransformerMixin
from sklearn.exceptions import NotFittedError
from typin... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | ```suggestion
X = X.drop(columns=self.invariant_cols)
```
The current option raises a FutureWarning in current Pandas versions. And inplace is usually frowned upon. Also this is one line shorter, and a lot faster. | jona-sassenhagen | 120 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/utils.py | """A collection of shared utilities for all encoders, not intended for external use."""
import pandas as pd
import numpy as np
from scipy.sparse import csr_matrix
__author__ = 'willmcginnis'
def convert_cols_to_list(cols):
if isinstance(cols, pd.Series):
return cols.tolist()
elif isinstance(cols, np... | """A collection of shared utilities for all encoders, not intended for external use."""
from abc import abstractmethod
from enum import Enum, auto
import pandas as pd
import numpy as np
import sklearn.base
from sklearn.base import BaseEstimator, TransformerMixin
from sklearn.exceptions import NotFittedError
from typin... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | ```suggestion
for x in self.invariant_cols:
self.feature_names.remove(x)
```
don't use a list comp for its side effects | jona-sassenhagen | 121 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/utils.py | """A collection of shared utilities for all encoders, not intended for external use."""
import pandas as pd
import numpy as np
from scipy.sparse import csr_matrix
__author__ = 'willmcginnis'
def convert_cols_to_list(cols):
if isinstance(cols, pd.Series):
return cols.tolist()
elif isinstance(cols, np... | """A collection of shared utilities for all encoders, not intended for external use."""
from abc import abstractmethod
from enum import Enum, auto
import pandas as pd
import numpy as np
import sklearn.base
from sklearn.base import BaseEstimator, TransformerMixin
from sklearn.exceptions import NotFittedError
from typin... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | What's the end purpose of the `EncodingRelation`s? | bmreiniger | 122 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/utils.py | """A collection of shared utilities for all encoders, not intended for external use."""
import pandas as pd
import numpy as np
from scipy.sparse import csr_matrix
__author__ = 'willmcginnis'
def convert_cols_to_list(cols):
if isinstance(cols, pd.Series):
return cols.tolist()
elif isinstance(cols, np... | """A collection of shared utilities for all encoders, not intended for external use."""
from abc import abstractmethod
from enum import Enum, auto
import pandas as pd
import numpy as np
import sklearn.base
from sklearn.base import BaseEstimator, TransformerMixin
from sklearn.exceptions import NotFittedError
from typin... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | With both supervised and unsupervised as classes, I think the tag `supervised_encoder` is probably better removed and replaced with an inheritance check. In the test files then, maybe just create at the top a sublist of `__all__` that pass that (instead of doing the check inside the loops of the test methods themselve... | bmreiniger | 123 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/utils.py | """A collection of shared utilities for all encoders, not intended for external use."""
import pandas as pd
import numpy as np
from scipy.sparse import csr_matrix
__author__ = 'willmcginnis'
def convert_cols_to_list(cols):
if isinstance(cols, pd.Series):
return cols.tolist()
elif isinstance(cols, np... | """A collection of shared utilities for all encoders, not intended for external use."""
from abc import abstractmethod
from enum import Enum, auto
import pandas as pd
import numpy as np
import sklearn.base
from sklearn.base import BaseEstimator, TransformerMixin
from sklearn.exceptions import NotFittedError
from typin... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | I introduced it in the first place to add the output shape of the transform method dynamically in the docstring, but didn't get it to work since docstrings are not meant to be dynamic (I guess).
I then decided to keep it since it could be useful metadata (e.g. you could add a test that the expected number of output ... | PaulWestenthanner | 124 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/utils.py | """A collection of shared utilities for all encoders, not intended for external use."""
import pandas as pd
import numpy as np
from scipy.sparse import csr_matrix
__author__ = 'willmcginnis'
def convert_cols_to_list(cols):
if isinstance(cols, pd.Series):
return cols.tolist()
elif isinstance(cols, np... | """A collection of shared utilities for all encoders, not intended for external use."""
from abc import abstractmethod
from enum import Enum, auto
import pandas as pd
import numpy as np
import sklearn.base
from sklearn.base import BaseEstimator, TransformerMixin
from sklearn.exceptions import NotFittedError
from typin... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | which could be actually a useful test by the way since when working on this I discovered that there might be problems with e.g. the backward difference encoder. This adds some intercept column and I guess if you encode two columns the latter intercept column overwrites the first one. But I'll have to check. If this tur... | PaulWestenthanner | 125 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/utils.py | """A collection of shared utilities for all encoders, not intended for external use."""
import pandas as pd
import numpy as np
from scipy.sparse import csr_matrix
__author__ = 'willmcginnis'
def convert_cols_to_list(cols):
if isinstance(cols, pd.Series):
return cols.tolist()
elif isinstance(cols, np... | """A collection of shared utilities for all encoders, not intended for external use."""
from abc import abstractmethod
from enum import Enum, auto
import pandas as pd
import numpy as np
import sklearn.base
from sklearn.base import BaseEstimator, TransformerMixin
from sklearn.exceptions import NotFittedError
from typin... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | I think both options are equivalent and more or less depend on taste, is it? Or is there any real difference that I'm missing? I'm also happy with either, not having a strong preference for one or the other | PaulWestenthanner | 126 |
scikit-learn-contrib/category_encoders | 325 | Refactor/base class | ## Proposed Changes
### Streamline Encoders
By introducing a `BaseEncoder` and a `Un/SupervisedTransformerMixin` that almost (c.f. below) all encoders inherit from the code is much more streamlined. This removes the boilerplate when implementing new encoders and hence makes it easier for new contributor.
### ... | null | 2021-11-28 12:23:41+00:00 | 2022-06-02 12:41:15+00:00 | category_encoders/utils.py | """A collection of shared utilities for all encoders, not intended for external use."""
import pandas as pd
import numpy as np
from scipy.sparse import csr_matrix
__author__ = 'willmcginnis'
def convert_cols_to_list(cols):
if isinstance(cols, pd.Series):
return cols.tolist()
elif isinstance(cols, np... | """A collection of shared utilities for all encoders, not intended for external use."""
from abc import abstractmethod
from enum import Enum, auto
import pandas as pd
import numpy as np
import sklearn.base
from sklearn.base import BaseEstimator, TransformerMixin
from sklearn.exceptions import NotFittedError
from typin... | PaulWestenthanner | a18cb64a81310a5e515c7b21255597b4dfb29b86 | 2e3282239ade4dfff362e655be0f65fe0d0270e9 | I think you're right that they're equivalent, I just think inheritance is simpler to understand, being base python instead of an sklearn mechanism. | bmreiniger | 127 |
scikit-learn-contrib/category_encoders | 322 | Fix ohe nan col | Fixes #295
## Proposed Changes
Prevents a column for missing values from being added in OneHotEncoder when handle_missing="error". Does this by preventing the underlying OrdinalEncoder from producing the mapping NaN->-2, by setting _its_ handle_missing to "error" as well.
Also patches an incidental bug in One... | null | 2021-10-29 02:20:47+00:00 | 2021-11-03 17:09:36+00:00 | category_encoders/one_hot.py | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | bmreiniger | cc0c4b9ab66a52979b37f791836bea1241046b8c | 0bcb96b6a505c9cec7c473578471491eab78b4eb | very nice catch!! | PaulWestenthanner | 128 |
scikit-learn-contrib/category_encoders | 322 | Fix ohe nan col | Fixes #295
## Proposed Changes
Prevents a column for missing values from being added in OneHotEncoder when handle_missing="error". Does this by preventing the underlying OrdinalEncoder from producing the mapping NaN->-2, by setting _its_ handle_missing to "error" as well.
Also patches an incidental bug in One... | null | 2021-10-29 02:20:47+00:00 | 2021-11-03 17:09:36+00:00 | category_encoders/one_hot.py | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | bmreiniger | cc0c4b9ab66a52979b37f791836bea1241046b8c | 0bcb96b6a505c9cec7c473578471491eab78b4eb | I think that's still not quite correct for the other options for handle missing:
for `return_nan` it still adds a column.
```python
df = pd.DataFrame([(1, "foo"), (2, "bar"), (4, None)], columns=["a", "b"])
encoder = OneHotEncoder(cols=["b"], handle_missing="return_nan", use_cat_names=True)
encoder.fit_transfor... | PaulWestenthanner | 129 |
scikit-learn-contrib/category_encoders | 322 | Fix ohe nan col | Fixes #295
## Proposed Changes
Prevents a column for missing values from being added in OneHotEncoder when handle_missing="error". Does this by preventing the underlying OrdinalEncoder from producing the mapping NaN->-2, by setting _its_ handle_missing to "error" as well.
Also patches an incidental bug in One... | null | 2021-10-29 02:20:47+00:00 | 2021-11-03 17:09:36+00:00 | category_encoders/one_hot.py | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | bmreiniger | cc0c4b9ab66a52979b37f791836bea1241046b8c | 0bcb96b6a505c9cec7c473578471491eab78b4eb | ~Maybe I don't understand what `indicator` is supposed to do: there's a test `test_HandleMissingIndicator_HaveNoNan_ExpectSecondColumn` that looks like what I expect `value` to do, not `indicator`.~
~Oh, maybe it's the unknown `value` that's spawning that extra column in that test.~
~Wait, no, I don't think I know ... | bmreiniger | 130 |
scikit-learn-contrib/category_encoders | 322 | Fix ohe nan col | Fixes #295
## Proposed Changes
Prevents a column for missing values from being added in OneHotEncoder when handle_missing="error". Does this by preventing the underlying OrdinalEncoder from producing the mapping NaN->-2, by setting _its_ handle_missing to "error" as well.
Also patches an incidental bug in One... | null | 2021-10-29 02:20:47+00:00 | 2021-11-03 17:09:36+00:00 | category_encoders/one_hot.py | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | bmreiniger | cc0c4b9ab66a52979b37f791836bea1241046b8c | 0bcb96b6a505c9cec7c473578471491eab78b4eb | OK, I think I have `handle_missing='return_nan'` working properly.
`handle_missing='indicator'` seems better off always generating the column, and that's what the code does and the tests test for. I'm in favor of just updating the docs to remove the warning (and explaining more what each option does). | bmreiniger | 131 |
scikit-learn-contrib/category_encoders | 322 | Fix ohe nan col | Fixes #295
## Proposed Changes
Prevents a column for missing values from being added in OneHotEncoder when handle_missing="error". Does this by preventing the underlying OrdinalEncoder from producing the mapping NaN->-2, by setting _its_ handle_missing to "error" as well.
Also patches an incidental bug in One... | null | 2021-10-29 02:20:47+00:00 | 2021-11-03 17:09:36+00:00 | category_encoders/one_hot.py | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | bmreiniger | cc0c4b9ab66a52979b37f791836bea1241046b8c | 0bcb96b6a505c9cec7c473578471491eab78b4eb | I agree with you. I also was wrong on what the `indicator` option is doing. The behaviour you implemented i.e. always create the new column in fit and then map the missing values to that option seems correct. I'd also welcome a better documentation on the available options. | PaulWestenthanner | 132 |
scikit-learn-contrib/category_encoders | 322 | Fix ohe nan col | Fixes #295
## Proposed Changes
Prevents a column for missing values from being added in OneHotEncoder when handle_missing="error". Does this by preventing the underlying OrdinalEncoder from producing the mapping NaN->-2, by setting _its_ handle_missing to "error" as well.
Also patches an incidental bug in One... | null | 2021-10-29 02:20:47+00:00 | 2021-11-03 17:09:36+00:00 | category_encoders/one_hot.py | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | bmreiniger | cc0c4b9ab66a52979b37f791836bea1241046b8c | 0bcb96b6a505c9cec7c473578471491eab78b4eb | maybe you can just use a dict here?
```
oe_handle_missing_mapping = {"error": "error", "return_nan": "return_nan", ... }
```
that way we'd also get a key error in case the option is invalid | PaulWestenthanner | 133 |
scikit-learn-contrib/category_encoders | 322 | Fix ohe nan col | Fixes #295
## Proposed Changes
Prevents a column for missing values from being added in OneHotEncoder when handle_missing="error". Does this by preventing the underlying OrdinalEncoder from producing the mapping NaN->-2, by setting _its_ handle_missing to "error" as well.
Also patches an incidental bug in One... | null | 2021-10-29 02:20:47+00:00 | 2021-11-03 17:09:36+00:00 | category_encoders/one_hot.py | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | """One-hot or dummy coding"""
import numpy as np
import pandas as pd
import warnings
from sklearn.base import BaseEstimator, TransformerMixin
from category_encoders.ordinal import OrdinalEncoder
import category_encoders.utils as util
__author__ = 'willmcginnis'
class OneHotEncoder(BaseEstimator, TransformerMixin):
... | bmreiniger | cc0c4b9ab66a52979b37f791836bea1241046b8c | 0bcb96b6a505c9cec7c473578471491eab78b4eb | `append_nan_to_index` would be a less hacky name. | PaulWestenthanner | 134 |
scikit-learn-contrib/category_encoders | 320 | Check array index fix | Closes #280.
Fixes #272, probably also #290, and supersedes #304.
## Proposed Changes
Replaces consecutive calls to `convert_input` (on `X`) and `convert_input_vector` (on `y`) by a single `convert_inputs` to ensure that the indexes of the results match. This is necessary for proper functioning of encoders that g... | null | 2021-10-24 21:33:05+00:00 | 2021-10-29 15:40:38+00:00 | category_encoders/utils.py | """A collection of shared utilities for all encoders, not intended for external use."""
import pandas as pd
import numpy as np
from scipy.sparse.csr import csr_matrix
__author__ = 'willmcginnis'
def convert_cols_to_list(cols):
if isinstance(cols, pd.Series):
return cols.tolist()
elif isinstance(cols... | """A collection of shared utilities for all encoders, not intended for external use."""
import pandas as pd
import numpy as np
from scipy.sparse.csr import csr_matrix
__author__ = 'willmcginnis'
def convert_cols_to_list(cols):
if isinstance(cols, pd.Series):
return cols.tolist()
elif isinstance(cols... | bmreiniger | 866bf143fb71db0de60d32e608393c1a3b8a71a7 | cc0c4b9ab66a52979b37f791836bea1241046b8c | we still support python 3.5 (although we should probably change that). However, because of this we cannot work with f-strings at the moment | PaulWestenthanner | 135 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.