dataforge-ml 0.11.0__tar.gz → 1.0.0__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/PKG-INFO +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/pyproject.toml +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/__init__.py +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/config.py +3 -3
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/__init__.py +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_base.py +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_boolean_profiler.py +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_correlation_profiler.py +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_tabular.py +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_target_profiler.py +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_text_profiler.py +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_type_detector.py +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/structural.py +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml.egg-info/PKG-INFO +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml.egg-info/SOURCES.txt +1 -1
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/LICENSE +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/README.md +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/setup.cfg +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/models/__init__.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/models/_data_structure.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/models/_data_types.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_boolean_config.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_categorical.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_categorical_config.py +0 -0
- /dataforge_ml-0.11.0/src/dataforge_ml/profiling/config.py → /dataforge_ml-1.0.0/src/dataforge_ml/profiling/_config.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_correlation_config.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_datetime_config.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_datetime_profiler.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_missingness_config.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_missingness_profiler.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_null_detection.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_numeric_config.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_numeric_profiler.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_target_config.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_text_config.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/splitting/__init__.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/splitting/_config.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/splitting/_splitter.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/utils/__init__.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/utils/data_loader.py +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml.egg-info/dependency_links.txt +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml.egg-info/requires.txt +0 -0
- {dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml.egg-info/top_level.txt +0 -0
|
@@ -6,7 +6,7 @@ from enum import StrEnum
|
|
|
6
6
|
from typing import TYPE_CHECKING, Union
|
|
7
7
|
|
|
8
8
|
if TYPE_CHECKING:
|
|
9
|
-
from dataforge_ml.profiling.
|
|
9
|
+
from dataforge_ml.profiling._config import ProfileConfig
|
|
10
10
|
|
|
11
11
|
|
|
12
12
|
class SemanticType(StrEnum):
|
|
@@ -32,7 +32,7 @@ class PipelinePhase(StrEnum):
|
|
|
32
32
|
|
|
33
33
|
|
|
34
34
|
def _default_profile_config() -> ProfileConfig:
|
|
35
|
-
from dataforge_ml.profiling.
|
|
35
|
+
from dataforge_ml.profiling._config import ProfileConfig
|
|
36
36
|
return ProfileConfig()
|
|
37
37
|
|
|
38
38
|
|
|
@@ -111,7 +111,7 @@ class PipelineConfig:
|
|
|
111
111
|
|
|
112
112
|
@classmethod
|
|
113
113
|
def from_dict(cls, data: dict) -> PipelineConfig:
|
|
114
|
-
from dataforge_ml.profiling.
|
|
114
|
+
from dataforge_ml.profiling._config import ProfileConfig
|
|
115
115
|
return cls(
|
|
116
116
|
exclude_columns=list(data.get("exclude_columns", [])),
|
|
117
117
|
phase_exclusions={
|
|
@@ -22,7 +22,7 @@ from __future__ import annotations
|
|
|
22
22
|
import polars as pl
|
|
23
23
|
|
|
24
24
|
from ._base import ColumnBatchProfiler
|
|
25
|
-
from .
|
|
25
|
+
from ._config import BooleanStats
|
|
26
26
|
from ._boolean_config import BooleanProfileResult
|
|
27
27
|
from ..models._data_types import _INT_DTYPES
|
|
28
28
|
|
{dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_correlation_profiler.py
RENAMED
|
@@ -42,7 +42,7 @@ from typing import Optional
|
|
|
42
42
|
import polars as pl
|
|
43
43
|
|
|
44
44
|
from ._base import DatasetLevelProfiler
|
|
45
|
-
from .
|
|
45
|
+
from ._config import ProfileConfig
|
|
46
46
|
from ._correlation_config import (
|
|
47
47
|
CategoricalTargetCorrelation,
|
|
48
48
|
CorrelationPair,
|
|
@@ -54,7 +54,7 @@ from __future__ import annotations
|
|
|
54
54
|
import polars as pl
|
|
55
55
|
|
|
56
56
|
from ._base import ColumnBatchProfiler
|
|
57
|
-
from .
|
|
57
|
+
from ._config import TextStats
|
|
58
58
|
from ._text_config import TextProfileResult
|
|
59
59
|
|
|
60
60
|
# Regex that counts non-whitespace token runs — used with str.count_matches.
|
|
@@ -20,7 +20,7 @@ from typing import TYPE_CHECKING
|
|
|
20
20
|
|
|
21
21
|
import polars as pl
|
|
22
22
|
|
|
23
|
-
from .
|
|
23
|
+
from ._config import ColumnTypeInfo, NumericKind, TypeFlag, SemanticType
|
|
24
24
|
from ..models._data_types import _INT_DTYPES, _NUMERIC_DTYPES
|
|
25
25
|
|
|
26
26
|
if TYPE_CHECKING:
|
|
@@ -35,7 +35,7 @@ from ._target_profiler import TargetProfiler
|
|
|
35
35
|
from ._correlation_profiler import CorrelationProfiler
|
|
36
36
|
from ._type_detector import TypeDetector
|
|
37
37
|
from ..config import PipelineConfig, PipelinePhase, SemanticType, Modality
|
|
38
|
-
from .
|
|
38
|
+
from ._config import (
|
|
39
39
|
ColumnProfile,
|
|
40
40
|
StructuralProfileResult,
|
|
41
41
|
RowMissingnessDistribution,
|
|
@@ -17,6 +17,7 @@ src/dataforge_ml/profiling/_boolean_config.py
|
|
|
17
17
|
src/dataforge_ml/profiling/_boolean_profiler.py
|
|
18
18
|
src/dataforge_ml/profiling/_categorical.py
|
|
19
19
|
src/dataforge_ml/profiling/_categorical_config.py
|
|
20
|
+
src/dataforge_ml/profiling/_config.py
|
|
20
21
|
src/dataforge_ml/profiling/_correlation_config.py
|
|
21
22
|
src/dataforge_ml/profiling/_correlation_profiler.py
|
|
22
23
|
src/dataforge_ml/profiling/_datetime_config.py
|
|
@@ -32,7 +33,6 @@ src/dataforge_ml/profiling/_target_profiler.py
|
|
|
32
33
|
src/dataforge_ml/profiling/_text_config.py
|
|
33
34
|
src/dataforge_ml/profiling/_text_profiler.py
|
|
34
35
|
src/dataforge_ml/profiling/_type_detector.py
|
|
35
|
-
src/dataforge_ml/profiling/config.py
|
|
36
36
|
src/dataforge_ml/profiling/structural.py
|
|
37
37
|
src/dataforge_ml/splitting/__init__.py
|
|
38
38
|
src/dataforge_ml/splitting/_config.py
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
{dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_categorical_config.py
RENAMED
|
File without changes
|
|
File without changes
|
{dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_correlation_config.py
RENAMED
|
File without changes
|
|
File without changes
|
|
File without changes
|
{dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_missingness_config.py
RENAMED
|
File without changes
|
{dataforge_ml-0.11.0 → dataforge_ml-1.0.0}/src/dataforge_ml/profiling/_missingness_profiler.py
RENAMED
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|