upgini 1.2.121a2__tar.gz → 1.2.121a3__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.
Files changed (82) hide show
  1. {upgini-1.2.121a2 → upgini-1.2.121a3}/PKG-INFO +1 -1
  2. upgini-1.2.121a3/src/upgini/__about__.py +1 -0
  3. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/features_validator.py +13 -6
  4. upgini-1.2.121a2/src/upgini/__about__.py +0 -1
  5. {upgini-1.2.121a2 → upgini-1.2.121a3}/.gitignore +0 -0
  6. {upgini-1.2.121a2 → upgini-1.2.121a3}/LICENSE +0 -0
  7. {upgini-1.2.121a2 → upgini-1.2.121a3}/README.md +0 -0
  8. {upgini-1.2.121a2 → upgini-1.2.121a3}/pyproject.toml +0 -0
  9. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/__init__.py +0 -0
  10. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/ads.py +0 -0
  11. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/ads_management/__init__.py +0 -0
  12. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/ads_management/ads_manager.py +0 -0
  13. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/__init__.py +0 -0
  14. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/all_operators.py +0 -0
  15. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/binary.py +0 -0
  16. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/date.py +0 -0
  17. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/feature.py +0 -0
  18. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/groupby.py +0 -0
  19. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/operator.py +0 -0
  20. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/timeseries/__init__.py +0 -0
  21. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/timeseries/base.py +0 -0
  22. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/timeseries/cross.py +0 -0
  23. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/timeseries/delta.py +0 -0
  24. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/timeseries/lag.py +0 -0
  25. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/timeseries/roll.py +0 -0
  26. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/timeseries/trend.py +0 -0
  27. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/timeseries/volatility.py +0 -0
  28. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/unary.py +0 -0
  29. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/utils.py +0 -0
  30. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/autofe/vector.py +0 -0
  31. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/data_source/__init__.py +0 -0
  32. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/data_source/data_source_publisher.py +0 -0
  33. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/dataset.py +0 -0
  34. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/errors.py +0 -0
  35. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/features_enricher.py +0 -0
  36. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/http.py +0 -0
  37. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/mdc/__init__.py +0 -0
  38. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/mdc/context.py +0 -0
  39. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/metadata.py +0 -0
  40. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/metrics.py +0 -0
  41. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/normalizer/__init__.py +0 -0
  42. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/normalizer/normalize_utils.py +0 -0
  43. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/resource_bundle/__init__.py +0 -0
  44. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/resource_bundle/exceptions.py +0 -0
  45. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/resource_bundle/strings.properties +0 -0
  46. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/resource_bundle/strings_widget.properties +0 -0
  47. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/sampler/__init__.py +0 -0
  48. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/sampler/base.py +0 -0
  49. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/sampler/random_under_sampler.py +0 -0
  50. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/sampler/utils.py +0 -0
  51. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/search_task.py +0 -0
  52. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/spinner.py +0 -0
  53. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/Roboto-Regular.ttf +0 -0
  54. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/__init__.py +0 -0
  55. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/base_search_key_detector.py +0 -0
  56. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/blocked_time_series.py +0 -0
  57. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/config.py +0 -0
  58. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/country_utils.py +0 -0
  59. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/custom_loss_utils.py +0 -0
  60. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/cv_utils.py +0 -0
  61. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/datetime_utils.py +0 -0
  62. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/deduplicate_utils.py +0 -0
  63. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/display_utils.py +0 -0
  64. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/email_utils.py +0 -0
  65. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/fallback_progress_bar.py +0 -0
  66. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/feature_info.py +0 -0
  67. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/format.py +0 -0
  68. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/hash_utils.py +0 -0
  69. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/ip_utils.py +0 -0
  70. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/mstats.py +0 -0
  71. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/phone_utils.py +0 -0
  72. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/postal_code_utils.py +0 -0
  73. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/progress_bar.py +0 -0
  74. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/psi.py +0 -0
  75. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/sample_utils.py +0 -0
  76. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/sklearn_ext.py +0 -0
  77. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/sort.py +0 -0
  78. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/target_utils.py +0 -0
  79. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/track_info.py +0 -0
  80. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/ts_utils.py +0 -0
  81. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/utils/warning_counter.py +0 -0
  82. {upgini-1.2.121a2 → upgini-1.2.121a3}/src/upgini/version_validator.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.3
2
2
  Name: upgini
3
- Version: 1.2.121a2
3
+ Version: 1.2.121a3
4
4
  Summary: Intelligent data search & enrichment for Machine Learning
5
5
  Project-URL: Bug Reports, https://github.com/upgini/upgini/issues
6
6
  Project-URL: Homepage, https://upgini.com/
@@ -0,0 +1 @@
1
+ __version__ = "1.2.121a3"
@@ -100,18 +100,25 @@ class FeaturesValidator:
100
100
  @staticmethod
101
101
  def is_one_hot_encoded(series: pd.Series) -> bool:
102
102
  try:
103
- # Column contains only 0 and 1 (as strings or numbers)
104
- series = series.astype(float)
105
- if set(series.unique()) != {0.0, 1.0}:
103
+ # First, handle string representations of True/False
104
+ series_copy = series.copy()
105
+ if series_copy.dtype == "object" or series_copy.dtype == "string":
106
+ # Convert string representations of boolean values to numeric
107
+ series_copy = series_copy.astype(str).str.strip().str.lower()
108
+ series_copy = series_copy.replace({"true": "1", "false": "0"})
109
+
110
+ # Column contains only 0 and 1 (as strings or numbers or booleans)
111
+ series_copy = series_copy.astype(float)
112
+ if set(series_copy.unique()) != {0.0, 1.0}:
106
113
  return False
107
114
 
108
- series = series.astype(int)
115
+ series_copy = series_copy.astype(int)
109
116
 
110
117
  # Column doesn't contain any NaN, np.NaN, space, null, etc.
111
- if not (series.isin([0, 1])).all():
118
+ if not (series_copy.isin([0, 1])).all():
112
119
  return False
113
120
 
114
- vc = series.value_counts()
121
+ vc = series_copy.value_counts()
115
122
  # Column should contain both 0 and 1
116
123
  if len(vc) != 2:
117
124
  return False
@@ -1 +0,0 @@
1
- __version__ = "1.2.121a2"
File without changes
File without changes
File without changes
File without changes
File without changes