dataspoof-mlpilot 2.1.0__tar.gz → 2.2.0__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.
Files changed (270) hide show
  1. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/CHANGELOG.md +262 -204
  2. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/Cargo.lock +1 -1
  3. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/Cargo.toml +1 -1
  4. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/PKG-INFO +1741 -1709
  5. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/README.md +1732 -1707
  6. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/architecture.md +32 -32
  7. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/deployment.md +72 -64
  8. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/media.md +48 -32
  9. dataspoof_mlpilot-2.2.0/docs/guide/quantization.md +143 -0
  10. dataspoof_mlpilot-2.2.0/docs/guide/vision.md +155 -0
  11. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/index.md +105 -105
  12. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/installation.md +90 -89
  13. dataspoof_mlpilot-2.2.0/docs/reference/compression.md +5 -0
  14. dataspoof_mlpilot-2.2.0/examples/titanic_walkthrough.ipynb +1313 -0
  15. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mkdocs.yml +141 -138
  16. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/__init__.py +63 -60
  17. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/automl/automl.py +630 -630
  18. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/automl/models.py +16 -1
  19. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/cicd.py +437 -396
  20. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/cleaners.py +3 -2
  21. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/cleaning.py +1367 -1368
  22. dataspoof_mlpilot-2.2.0/mlpilot/compression.py +221 -0
  23. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/deploy/__init__.py +21 -18
  24. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/deploy/cli.py +299 -284
  25. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/deploy/export.py +317 -285
  26. dataspoof_mlpilot-2.2.0/mlpilot/deploy/quantize.py +341 -0
  27. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/features/text.py +156 -133
  28. dataspoof_mlpilot-2.2.0/mlpilot/media/__init__.py +45 -0
  29. dataspoof_mlpilot-2.2.0/mlpilot/media/benchmark.py +172 -0
  30. dataspoof_mlpilot-2.2.0/mlpilot/media/conditions.py +414 -0
  31. dataspoof_mlpilot-2.2.0/mlpilot/media/faces.py +343 -0
  32. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/media/image.py +365 -353
  33. dataspoof_mlpilot-2.2.0/mlpilot/media/plates.py +292 -0
  34. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/steps.py +930 -926
  35. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/pyproject.toml +234 -222
  36. dataspoof_mlpilot-2.2.0/tests/assets/astronaut.jpg +0 -0
  37. dataspoof_mlpilot-2.2.0/tests/assets/coffee.jpg +0 -0
  38. dataspoof_mlpilot-2.2.0/tests/test_quantization.py +244 -0
  39. dataspoof_mlpilot-2.2.0/tests/test_vision.py +273 -0
  40. dataspoof_mlpilot-2.1.0/mlpilot/media/__init__.py +0 -19
  41. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/.dockerignore +0 -0
  42. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/.github/workflows/CI.yml +0 -0
  43. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/.github/workflows/docs.yml +0 -0
  44. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/.gitignore +0 -0
  45. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/BUILD.md +0 -0
  46. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/CONTRIBUTING.md +0 -0
  47. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/SECURITY.md +0 -0
  48. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/helm/mlpilot-hub/Chart.yaml +0 -0
  49. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/helm/mlpilot-hub/templates/_helpers.tpl +0 -0
  50. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/helm/mlpilot-hub/templates/deployment.yaml +0 -0
  51. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/helm/mlpilot-hub/templates/ingress.yaml +0 -0
  52. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/helm/mlpilot-hub/templates/pdb.yaml +0 -0
  53. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/helm/mlpilot-hub/templates/pvc.yaml +0 -0
  54. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/helm/mlpilot-hub/templates/secret.yaml +0 -0
  55. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/helm/mlpilot-hub/templates/service.yaml +0 -0
  56. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/helm/mlpilot-hub/templates/serviceaccount.yaml +0 -0
  57. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/helm/mlpilot-hub/values.yaml +0 -0
  58. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/hub/Dockerfile +0 -0
  59. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/hub/docker-compose.yml +0 -0
  60. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/monitoring/docker-compose.yml +0 -0
  61. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/monitoring/grafana/dashboards/mlpilot-model.json +0 -0
  62. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/monitoring/grafana/provisioning/dashboards/mlpilot.yml +0 -0
  63. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/monitoring/grafana/provisioning/datasources/prometheus.yml +0 -0
  64. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/monitoring/prometheus/alerts.yml +0 -0
  65. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/monitoring/prometheus/alerts_test.yml +0 -0
  66. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/monitoring/prometheus/prometheus.yml +0 -0
  67. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/opa/policies/mlpilot/authz.rego +0 -0
  68. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/opa/policies/mlpilot/config/data.json +0 -0
  69. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/opa/policies/mlpilot/policy_test.rego +0 -0
  70. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/opa/policies/mlpilot/promotion.rego +0 -0
  71. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/terraform/aws/main.tf +0 -0
  72. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/terraform/aws/outputs.tf +0 -0
  73. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/terraform/aws/variables.tf +0 -0
  74. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/terraform/aws-ecs/main.tf +0 -0
  75. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/terraform/aws-ecs/outputs.tf +0 -0
  76. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/terraform/aws-ecs/variables.tf +0 -0
  77. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/terraform/aws-sagemaker/main.tf +0 -0
  78. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/terraform/aws-sagemaker/outputs.tf +0 -0
  79. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/deploy/terraform/aws-sagemaker/variables.tf +0 -0
  80. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/assets/workflow.svg +0 -0
  81. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/changelog.md +0 -0
  82. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/contributing.md +0 -0
  83. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/anomaly.md +0 -0
  84. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/automl.md +0 -0
  85. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/aws.md +0 -0
  86. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/causal.md +0 -0
  87. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/cleaning.md +0 -0
  88. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/environments-cicd.md +0 -0
  89. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/evaluation.md +0 -0
  90. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/feature-store.md +0 -0
  91. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/features.md +0 -0
  92. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/forecasting.md +0 -0
  93. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/hub.md +0 -0
  94. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/ingestion.md +0 -0
  95. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/insights.md +0 -0
  96. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/integrations.md +0 -0
  97. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/lineage.md +0 -0
  98. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/model-security.md +0 -0
  99. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/monitoring.md +0 -0
  100. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/observability.md +0 -0
  101. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/quality.md +0 -0
  102. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/registry.md +0 -0
  103. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/secrets-policy.md +0 -0
  104. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/splitting.md +0 -0
  105. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/stats.md +0 -0
  106. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/guide/versioning.md +0 -0
  107. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/policies.md +0 -0
  108. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/anomaly.md +0 -0
  109. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/automl.md +0 -0
  110. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/causal.md +0 -0
  111. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/cicd.md +0 -0
  112. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/cleaners.md +0 -0
  113. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/core.md +0 -0
  114. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/deploy.md +0 -0
  115. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/environments.md +0 -0
  116. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/evaluation.md +0 -0
  117. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/feature_store.md +0 -0
  118. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/features.md +0 -0
  119. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/forecast.md +0 -0
  120. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/hub.md +0 -0
  121. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/ingestion.md +0 -0
  122. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/insights.md +0 -0
  123. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/integrations.md +0 -0
  124. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/lineage.md +0 -0
  125. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/media.md +0 -0
  126. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/monitoring.md +0 -0
  127. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/poisoning.md +0 -0
  128. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/quality.md +0 -0
  129. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/registry.md +0 -0
  130. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/robustness.md +0 -0
  131. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/scale.md +0 -0
  132. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/secretstore.md +0 -0
  133. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/security.md +0 -0
  134. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/splitting.md +0 -0
  135. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/stats.md +0 -0
  136. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/reference/versioning.md +0 -0
  137. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/security.md +0 -0
  138. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/docs/workflow.md +0 -0
  139. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/_compat.py +0 -0
  140. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/_detect.py +0 -0
  141. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/_rustcore.pyi +0 -0
  142. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/anomaly.py +0 -0
  143. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/automl/__init__.py +0 -0
  144. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/automl/_persist.py +0 -0
  145. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/automl/cluster.py +0 -0
  146. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/automl/frame.py +0 -0
  147. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/automl/tasks.py +0 -0
  148. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/causal.py +0 -0
  149. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/core/__init__.py +0 -0
  150. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/core/pipeline.py +0 -0
  151. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/core/registry.py +0 -0
  152. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/core/step.py +0 -0
  153. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/deploy/__main__.py +0 -0
  154. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/deploy/_names.py +0 -0
  155. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/deploy/asgi.py +0 -0
  156. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/deploy/aws.py +0 -0
  157. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/deploy/scoring.py +0 -0
  158. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/deploy/server.py +0 -0
  159. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/deploy/wsgi.py +0 -0
  160. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/environments.py +0 -0
  161. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/evaluation/__init__.py +0 -0
  162. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/evaluation/_slicing.py +0 -0
  163. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/evaluation/adapter.py +0 -0
  164. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/evaluation/card.py +0 -0
  165. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/evaluation/errors.py +0 -0
  166. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/evaluation/explain.py +0 -0
  167. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/evaluation/fairness.py +0 -0
  168. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/evaluation/metrics.py +0 -0
  169. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/feature_store.py +0 -0
  170. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/features/__init__.py +0 -0
  171. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/features/dates.py +0 -0
  172. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/features/defaults.py +0 -0
  173. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/features/encoding.py +0 -0
  174. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/features/generators.py +0 -0
  175. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/features/leakage.py +0 -0
  176. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/features/selection.py +0 -0
  177. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/forecast.py +0 -0
  178. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/hub/__init__.py +0 -0
  179. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/hub/app.py +0 -0
  180. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/hub/client.py +0 -0
  181. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/hub/db.py +0 -0
  182. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/hub/policy.py +0 -0
  183. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/hub/security.py +0 -0
  184. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/hub/storage.py +0 -0
  185. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/hub/ui.html +0 -0
  186. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/hub/ui.js +0 -0
  187. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/ingestion/__init__.py +0 -0
  188. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/ingestion/_util.py +0 -0
  189. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/ingestion/api.py +0 -0
  190. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/ingestion/core.py +0 -0
  191. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/ingestion/files.py +0 -0
  192. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/ingestion/incremental.py +0 -0
  193. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/ingestion/schema.py +0 -0
  194. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/ingestion/sql.py +0 -0
  195. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/ingestion/warehouse.py +0 -0
  196. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/insights/__init__.py +0 -0
  197. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/insights/finder.py +0 -0
  198. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/insights/narrate.py +0 -0
  199. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/insights/report.py +0 -0
  200. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/__init__.py +0 -0
  201. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/airflow.py +0 -0
  202. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/aws_glue.py +0 -0
  203. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/azure_data_factory.py +0 -0
  204. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/dagster.py +0 -0
  205. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/dvc.py +0 -0
  206. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/evidently.py +0 -0
  207. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/feast.py +0 -0
  208. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/kafka.py +0 -0
  209. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/mlflow.py +0 -0
  210. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/prefect.py +0 -0
  211. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/integrations/tasks.py +0 -0
  212. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/lineage.py +0 -0
  213. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/media/audio.py +0 -0
  214. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/monitoring/__init__.py +0 -0
  215. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/monitoring/drift.py +0 -0
  216. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/monitoring/estimate.py +0 -0
  217. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/monitoring/monitor.py +0 -0
  218. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/monitoring/ops.py +0 -0
  219. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/poisoning.py +0 -0
  220. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/py.typed +0 -0
  221. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/quality/__init__.py +0 -0
  222. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/quality/checks.py +0 -0
  223. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/quality/contracts.py +0 -0
  224. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/quality/pipeline_steps.py +0 -0
  225. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/quality/profile.py +0 -0
  226. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/quality/schema.py +0 -0
  227. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/registry/__init__.py +0 -0
  228. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/registry/_store.py +0 -0
  229. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/registry/models.py +0 -0
  230. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/registry/tracking.py +0 -0
  231. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/robustness.py +0 -0
  232. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/scale.py +0 -0
  233. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/secretstore.py +0 -0
  234. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/security.py +0 -0
  235. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/splitting.py +0 -0
  236. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/stats.py +0 -0
  237. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/mlpilot/versioning.py +0 -0
  238. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/src/clean.rs +0 -0
  239. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/src/lib.rs +0 -0
  240. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/src/stats.rs +0 -0
  241. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/src/text.rs +0 -0
  242. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_advanced.py +0 -0
  243. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_automl.py +0 -0
  244. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_aws_deploy.py +0 -0
  245. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_cicd.py +0 -0
  246. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_cleaners.py +0 -0
  247. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_core.py +0 -0
  248. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_data_versions_and_feature_store.py +0 -0
  249. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_deploy.py +0 -0
  250. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_docker_stacks.py +0 -0
  251. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_environments.py +0 -0
  252. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_evaluation.py +0 -0
  253. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_features.py +0 -0
  254. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_hub.py +0 -0
  255. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_ingestion.py +0 -0
  256. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_insights.py +0 -0
  257. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_label_free_monitoring.py +0 -0
  258. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_lineage_and_catalogs.py +0 -0
  259. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_media.py +0 -0
  260. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_model_security.py +0 -0
  261. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_monitoring.py +0 -0
  262. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_ops.py +0 -0
  263. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_quality.py +0 -0
  264. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_real_integrations.py +0 -0
  265. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_registry.py +0 -0
  266. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_rename_compat.py +0 -0
  267. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_secretstore.py +0 -0
  268. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_serving_feedback_reload.py +0 -0
  269. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_splitting.py +0 -0
  270. {dataspoof_mlpilot-2.1.0 → dataspoof_mlpilot-2.2.0}/tests/test_stats.py +0 -0
@@ -1,204 +1,262 @@
1
- # Changelog
2
-
3
- All notable changes are listed here. The project follows [Semantic Versioning](https://semver.org/)
4
- (see the [versioning policy](https://dataspoof.github.io/dataspoof-data-cleaning/policies/)).
5
-
6
- ## [2.1.0] — 2026-11-09
7
-
8
- ### Renamed: AutomatedCleaning is now MLPilot
9
- - **Names:**
10
- - the distribution is `dataspoof-mlpilot` (PyPI refused `mlpilot` as too similar to an existing project);
11
- - the import is `import mlpilot`, and the CLI is `mlpilot`;
12
- - `MLPilotWarning` replaces `AutomatedCleaningWarning`, which remains as an alias.
13
- - **Compatibility:**
14
- - `import automatedcleaning` and pickles that refer to `automatedcleaning.*` classes resolve to
15
- the same `mlpilot` modules, so models saved by AutomatedCleaning keep loading with the same
16
- classes;
17
- - the `AutomatedCleaning` project on PyPI gets a final 2.0.1 that installs `dataspoof-mlpilot`.
18
- - **Unchanged:** the `AC_*` environment variables, the `ac_*` Prometheus metrics and the saved-file
19
- formats (`dataspoof.*`).
20
- - **Renamed with the library:**
21
- - the Helm chart (`mlpilot-hub`), the OPA policy package (`data.mlpilot`) and the Prometheus job;
22
- - the Grafana dashboard file;
23
- - the MLflow tag prefix and the default OpenLineage namespace.
24
-
25
- ## [2.0.0] — 2026-11-09
26
-
27
- ### Breaking
28
- - **Lean core install.** `pip install AutomatedCleaning` now installs only the core
29
- dependencies. The heavy ones moved to extras:
30
- - `[nlp]`: nltk
31
- - `[dashboard]`: plotly, matplotlib, missingno, pyfiglet
32
- - `[pii]`: presidio
33
- - `[llm]`: anthropic, langchain-anthropic
34
-
35
- To get what 1.x installed, use `pip install "AutomatedCleaning[all]"`. A feature whose extra
36
- is missing raises an `ImportError` naming the extra to install.
37
- - `setup.py` was removed. The build is defined only in `pyproject.toml` (maturin).
38
-
39
- ### Changed
40
- - **No side effects at import.**
41
- - Importing the library no longer downloads NLTK data, changes the application's warning
42
- filters, or reconfigures logging.
43
- - NLTK corpora are downloaded on first use, and never when `AC_OFFLINE` is set. Text cleaning
44
- falls back to a built-in stopword list when NLTK isn't installed.
45
- - **Typed package.** It ships a `py.typed` marker and stubs for the Rust extension, and passes
46
- `mypy`.
47
- - **CI** now runs:
48
- - tests across Linux, macOS and Windows × Python 3.11–3.13 × Polars 1/2;
49
- - a core-only install test;
50
- - mypy;
51
- - pip-audit and cargo-audit.
52
-
53
- Releases publish with PyPI Trusted Publishing, signed build-provenance attestations and a
54
- CycloneDX SBOM.
55
- - Interactive AI corrections use `ANTHROPIC_API_KEY` when it is set, and the current Claude model.
56
-
57
- ### Added
58
- - **Hub, the multi-user platform server** (`automatedcleaning.hub`, `[server]` extra):
59
- - workspaces, with SSO through OIDC and identity-provider group mapping;
60
- - API tokens and service accounts;
61
- - four roles (viewer, data scientist, approver, admin);
62
- - runs and a model registry with immutable, checksummed artifacts on disk, S3, GCS or Azure;
63
- - two-person production approvals with policy gates;
64
- - monitoring results;
65
- - a hash-chained, verifiable and exportable audit log;
66
- - a web UI, the Python client `HubClient`, and `automatedcleaning hub serve | create-admin`.
67
- - **Hub deployment:** a Docker image with compose (plus PostgreSQL), a Helm chart, and a
68
- Terraform module for AWS (S3, RDS, Secrets Manager, IAM role, Helm release).
69
- - **Forecasting** (`Forecaster`):
70
- - multiple series and multi-step horizons;
71
- - rolling-origin backtests against seasonal-naive;
72
- - conformal intervals and hierarchical reconciliation.
73
- - **Causal inference** (`estimate_effect`): doubly-robust AIPW with overlap and balance
74
- diagnostics. **Uplift:** `UpliftModel` and `qini`.
75
- - **Anomaly detection** (`AutoAnomaly`): an ensemble of three detectors with plain-language
76
- explanations.
77
- - **AutoML:**
78
- - conformal `predict_interval` and `predict_set`;
79
- - `search="optuna"`, `imbalance="auto"`, `n_jobs`;
80
- - `predict_with_proba` and `set_inference_threads`.
81
- - **Serving:**
82
- - Prometheus `/metrics` and OpenTelemetry tracing;
83
- - canary and shadow rollouts, and `app_from_registry`;
84
- - SageMaker, Vertex AI and Azure ML routes and Docker targets;
85
- - an ASGI app and the `container` entry point;
86
- - single-row latency down from about 500 ms to about 50 ms.
87
- - **Security:**
88
- - Ed25519 model signing, verified before unpickling whenever `AC_MODEL_PUBLIC_KEY` is set;
89
- - `skops` export;
90
- - signing on registration with `register(..., sign_with=...)`.
91
- - **Monitoring:**
92
- - PagerDuty and Opsgenie alert handlers;
93
- - `join_labels` and `Monitor.check_labels` for labels that arrive later;
94
- - `sql_sink` and `Monitor.on_check`;
95
- - `SQLSource.insert` and `execute`.
96
- - **Scale:**
97
- - `transform_file` for files of any size;
98
- - Spark, Ray and Dask scoring;
99
- - `to_sql` SQL pushdown of fitted cleaning pipelines.
100
- - **Integrations:** Airflow operators, Prefect tasks and flow, Dagster assets, Kafka stream
101
- scoring, and plain `integrations.tasks`.
102
- - **MLflow** (`integrations.mlflow`, `[mlflow]` extra):
103
- - batched, idempotent export of runs and whole trackers;
104
- - AutoML models logged as pyfunc models with a signature, so they serve with `mlflow models serve`;
105
- - `export_registry` (stages become aliases) and `load_automl`.
106
- - **DVC** (`integrations.dvc`, `[dvc]` extra): `DVCRepo` (init, remotes, track, push and pull,
107
- read any revision, checkout, `log_to_run`) and `export_repository` for `DataRepository`
108
- history.
109
- - **Feast** (`integrations.feast`, `[feast]` extra): `export_to_feast` writes, applies and
110
- materializes a Feast repository.
111
- - **Data splitting** (`split_data`, `cv_folds`, `check_split`): stratified, grouped and time-ordered
112
- splits with gaps; every split is leak-checked (shared groups, time overlap, duplicate rows, unseen classes).
113
- - **Model security:** `robustness_report` (noise stability, a minimal adversarial attack,
114
- feature sensitivity, extreme inputs) and `poisoning_report` (confident-learning label errors,
115
- conflicting duplicates, backdoor triggers, outliers).
116
- - **Environments** (`Environments`, `environments.yml`): isolated dev / QA / prod registries;
117
- `release` enforces order, source stage, metric gates, four-eyes approvals and signatures, and
118
- records provenance on both sides. The registry and Hub gain a `qa` stage.
119
- - **Model CI/CD:** the CLI commands `train`, `gate`, `register`, `promote`, `release`, `latest`,
120
- `deploy-sagemaker` and `cicd init` (GitHub Actions / GitLab CI pipelines, checked with actionlint).
121
- - **Secrets** (`secretstore`): `vault://` (token, AppRole, Kubernetes; KV v1 and v2), `aws-sm://`,
122
- `env://` and `file://` references, resolved wherever credentials are read.
123
- - **Open Policy Agent for the Hub** (`hub.policy`): deny-overrides access rules, promotion rules
124
- and fail-closed behaviour, plus Rego policies with unit tests in `deploy/opa`.
125
- - **Serving:**
126
- - `POST /feedback` with prediction IDs, live-accuracy metrics and `join_feedback`;
127
- - hot reload (`watch=`, `POST /admin/reload`, registry-following);
128
- - live drift, estimated-performance and alert gauges (`monitor=`).
129
- - **Monitoring before labels:** `PerformanceEstimator` (CBPE, direct loss estimation),
130
- `domain_classifier_drift`, and `estimated_performance` alerts in `Monitor.check`.
131
- - **Grafana + Prometheus** (`deploy/monitoring`): a dashboard, 14 alert rules with `promtool`
132
- unit tests, and a compose stack.
133
- - **Lineage** (`lineage`): OpenLineage run events with schema, quality and version facets; HTTP
134
- and file transports; `emit_tracked_run`, `emit_dataset_version`, `LineageGraph`.
135
- - **Catalogs and orchestration:** AWS Glue (catalog tables, partitions, reads, job runs) and Azure
136
- Data Factory pipeline runs.
137
- - **AWS deployment** (`deploy.aws`): ECR push and blue/green SageMaker endpoints (serverless or
138
- instances, autoscaling, data capture, rollback alarms); Terraform for SageMaker and ECS Fargate.
139
- - **Evidently** (`integrations.evidently`): reports plus agreement with the built-in drift report.
140
- - **Tests against real infrastructure:** Prometheus, Grafana, Vault, OPA and Marquez in Docker
141
- (`AC_DOCKER_TESTS=1`); moto for AWS; dedicated CI jobs.
142
- - These three are tested against the real tools (MLflow server and model server, git + DVC
143
- with a remote, Feast), including in a dedicated CI job.
144
- - **Image and audio data** (`media`):
145
- - quality audits (corrupt files, exact and near duplicates, blur, exposure, clipping, silence,
146
- mixed formats);
147
- - preprocessors (EXIF-aware image resize and normalise; audio resample, trim and level);
148
- - augmentation;
149
- - features for AutoML (colour and edge-orientation histograms; MFCC and spectral features).
150
- - **Statistical testing** (`stats`):
151
- - automatic test selection with assumption checks, effect sizes and confidence intervals;
152
- - chi-square and Fisher's exact test, correlations;
153
- - A/B tests, power analysis, multiple-testing correction, `test_all`.
154
- - **Data version control** (`DataRepository`): content-addressed commits, tags, row-level
155
- diffs, checksum-verified checkout, links to experiment runs, garbage collection.
156
- - **Feature store** (`FeatureStore`, `Entity`, `FeatureView`): point-in-time-correct
157
- historical joins with TTL, an SQLite online store, `materialize`, schema guards.
158
- - **Documentation site** (MkDocs Material with an API reference), plus `SECURITY.md`,
159
- `CONTRIBUTING.md` and a versioning policy.
160
-
161
- ### Removed
162
- - `automatedcleaning/cleaning_update.py`: dead code that nothing imported.
163
-
164
- ## [1.14.0] — 2026-11-08
165
- - **Ingestion & connectors:**
166
- - files, globs, HTTP, and S3 / GCS / Azure;
167
- - any SQL database (bound parameters);
168
- - Snowflake and BigQuery;
169
- - REST APIs with five pagination styles;
170
- - reusable schema inference and incremental loads.
171
- - **Experiment tracking & model registry:** runs, dataset fingerprints, dev → staging →
172
- production promotion with metric gates, audit history and rollback.
173
- - **Deployment:**
174
- - batch scoring and a REST API (API keys, prediction log);
175
- - bundle, pickle and ONNX export;
176
- - Docker build contexts and a CLI.
177
- - **Monitoring:** data and prediction drift (PSI), accuracy decay, alerts and webhooks, and
178
- champion-vs-challenger retraining.
179
- - **Insights & reporting:** significance-tested findings, executive reports, and optional
180
- narratives written by Claude.
181
-
182
- ## [1.13.0] — 2026-11-06
183
- - **Feature engineering:**
184
- - dates, out-of-fold target encoding, interactions, aggregations, lags, text embeddings;
185
- - feature selection and leakage detection.
186
- - **AutoML:** task detection, time-budgeted model search, an ensemble, and AutoCluster.
187
- - **Evaluation & explainability:** metrics with confidence intervals, built-in SHAP, error
188
- analysis, fairness checks and model cards.
189
-
190
- ## [1.12.0] — 2026-11-06
191
- - Profiling & data quality: automated EDA, the 0–100 quality score, YAML data contracts, schema
192
- diffs and quality gates.
193
-
194
- ## [1.11.0] — 2026-11-05
195
- - Shared `Step` / `Pipeline` core and the stateful cleaning pipeline (JSON save/load, audit
196
- reports, schema checks).
197
-
198
- ## [1.10.0] and earlier — 2026-07
199
- - 1.10.0: 8 more data-quality checks in the dashboard.
200
- - 1.9.0: column pickers in the dashboard.
201
- - 1.8.x: the dashboard data-quality panel.
202
- - 1.7.0: the enterprise dashboard redesign.
203
- - 1.6.x: 43 additional cleaning steps and a dependency cleanup.
204
- - 1.5.x: the Rust compute backend.
1
+ # Changelog
2
+
3
+ All notable changes are listed here. The project follows [Semantic Versioning](https://semver.org/)
4
+ (see the [versioning policy](https://dataspoof.github.io/dataspoof-data-cleaning/policies/)).
5
+
6
+ ## [2.2.0] — 2026-11-12
7
+
8
+ ### Added: images taken in hard conditions
9
+ - **Weather and lighting** (`media.conditions`):
10
+ - detects fog, rain, snow, low light, glare, motion blur, sensor noise and JPEG damage
11
+ (each scored 0-1 and tagged at 0.5);
12
+ - `condition_report` shows which conditions a dataset covers and flags the rare or missing ones;
13
+ - `apply_condition` simulates all of them, plus shadow and cutout, at severity 1-5;
14
+ - `enhance` restores images: dehaze, low-light brightening, denoise, sharpen.
15
+ - **Faces** (`media.faces`, `[vision]` extra):
16
+ - detection;
17
+ - how much of a face is covered, and by what (hair, mask, sunglasses or object), in a per-image
18
+ `occlusion_report`;
19
+ - simulated coverings for training;
20
+ - blurring or pixelating for privacy (`anonymize`, `blur_regions`).
21
+ - **Licence plates** (`media.plates`, `[ocr]` extra):
22
+ - detection and OCR, with plate patterns (`IN`, `IN_BH`, `UK`, `EU`, `US`) that fix OCR
23
+ look-alikes;
24
+ - optional restoration before reading;
25
+ - `plate_report`, which pairs each read with the conditions in the image.
26
+ - **Robustness benchmark** (`media.benchmark`):
27
+ - `condition_benchmark` re-scores any image model per condition and severity, with face
28
+ coverings included;
29
+ - `condition_augment` builds training sets from the weakest conditions;
30
+ - ready-made predictors for AutoML image classifiers, plate reading and face counting.
31
+ - New extras: `vision` (OpenCV 4: OpenCV 5 no longer ships the detector models) and `ocr`
32
+ (RapidOCR; its models are bundled, so nothing is downloaded at run time).
33
+ - Image functions also accept numpy arrays, alongside PIL images and paths.
34
+
35
+ ### Added: quantization
36
+ - **Models** (`mlpilot.deploy.quantize`, `[onnx]` extra):
37
+ - `quantize_model` writes int8 and fp16 ONNX variants of a model and measures each on held-out
38
+ data (file size, speed, metric change and agreement with the original), then recommends
39
+ one;
40
+ - `export_model(..., format="onnx", quantize="int8" | "uint8" | "fp16")`, `quantize_onnx`
41
+ for any ONNX model, and a `mlpilot quantize` CLI command;
42
+ - models with nothing to quantize (tree ensembles, linear models, k-NN) are reported as
43
+ such rather than exported unchanged.
44
+ - **AutoML:** a new opt-in `mlp` model family (scikit-learn neural network), used only when
45
+ named in `include=`. It is the AutoML model family that quantization shrinks.
46
+ - **CI/CD:** a `quantization` gate in `gates.yml` (`mode`, `max_drop`, `min_agreement`,
47
+ `required`).
48
+ - **Embeddings:**
49
+ - `TextEmbeddings(dtype="float32" | "float16" | "int8")`;
50
+ - `quantize_vectors` / `dequantize_vectors` with a per-dimension int8 scale learned on
51
+ training data.
52
+ - **Tables:**
53
+ - `optimize_dtypes` and the `OptimizeDtypes` pipeline step shrink column types without
54
+ changing values (integers by range, floats only when they survive the round trip, text
55
+ to Categorical);
56
+ - the step learns the types on training data and never overflows on new data.
57
+
58
+ ### Fixed
59
+ - `save_cleaned_data(quantize=True)` no longer fails on integers above 2.1 billion (it cast
60
+ every `Int64` column to `Int32`).
61
+ - `ClipOutliers` no longer fails on small integer columns (`Int8`, `Int16`, `UInt8`, …)
62
+ whose learned fences lie outside the type's range.
63
+
64
+ ## [2.1.0] — 2026-11-09
65
+
66
+ ### Renamed: AutomatedCleaning is now MLPilot
67
+ - **Names:**
68
+ - the distribution is `dataspoof-mlpilot` (PyPI refused `mlpilot` as too similar to an existing project);
69
+ - the import is `import mlpilot`, and the CLI is `mlpilot`;
70
+ - `MLPilotWarning` replaces `AutomatedCleaningWarning`, which remains as an alias.
71
+ - **Compatibility:**
72
+ - `import automatedcleaning` and pickles that refer to `automatedcleaning.*` classes resolve to
73
+ the same `mlpilot` modules, so models saved by AutomatedCleaning keep loading with the same
74
+ classes;
75
+ - the `AutomatedCleaning` project on PyPI gets a final 2.0.1 that installs `dataspoof-mlpilot`.
76
+ - **Unchanged:** the `AC_*` environment variables, the `ac_*` Prometheus metrics and the saved-file
77
+ formats (`dataspoof.*`).
78
+ - **Renamed with the library:**
79
+ - the Helm chart (`mlpilot-hub`), the OPA policy package (`data.mlpilot`) and the Prometheus job;
80
+ - the Grafana dashboard file;
81
+ - the MLflow tag prefix and the default OpenLineage namespace.
82
+
83
+ ## [2.0.0] — 2026-11-09
84
+
85
+ ### Breaking
86
+ - **Lean core install.** `pip install AutomatedCleaning` now installs only the core
87
+ dependencies. The heavy ones moved to extras:
88
+ - `[nlp]`: nltk
89
+ - `[dashboard]`: plotly, matplotlib, missingno, pyfiglet
90
+ - `[pii]`: presidio
91
+ - `[llm]`: anthropic, langchain-anthropic
92
+
93
+ To get what 1.x installed, use `pip install "AutomatedCleaning[all]"`. A feature whose extra
94
+ is missing raises an `ImportError` naming the extra to install.
95
+ - `setup.py` was removed. The build is defined only in `pyproject.toml` (maturin).
96
+
97
+ ### Changed
98
+ - **No side effects at import.**
99
+ - Importing the library no longer downloads NLTK data, changes the application's warning
100
+ filters, or reconfigures logging.
101
+ - NLTK corpora are downloaded on first use, and never when `AC_OFFLINE` is set. Text cleaning
102
+ falls back to a built-in stopword list when NLTK isn't installed.
103
+ - **Typed package.** It ships a `py.typed` marker and stubs for the Rust extension, and passes
104
+ `mypy`.
105
+ - **CI** now runs:
106
+ - tests across Linux, macOS and Windows × Python 3.11–3.13 × Polars 1/2;
107
+ - a core-only install test;
108
+ - mypy;
109
+ - pip-audit and cargo-audit.
110
+
111
+ Releases publish with PyPI Trusted Publishing, signed build-provenance attestations and a
112
+ CycloneDX SBOM.
113
+ - Interactive AI corrections use `ANTHROPIC_API_KEY` when it is set, and the current Claude model.
114
+
115
+ ### Added
116
+ - **Hub, the multi-user platform server** (`automatedcleaning.hub`, `[server]` extra):
117
+ - workspaces, with SSO through OIDC and identity-provider group mapping;
118
+ - API tokens and service accounts;
119
+ - four roles (viewer, data scientist, approver, admin);
120
+ - runs and a model registry with immutable, checksummed artifacts on disk, S3, GCS or Azure;
121
+ - two-person production approvals with policy gates;
122
+ - monitoring results;
123
+ - a hash-chained, verifiable and exportable audit log;
124
+ - a web UI, the Python client `HubClient`, and `automatedcleaning hub serve | create-admin`.
125
+ - **Hub deployment:** a Docker image with compose (plus PostgreSQL), a Helm chart, and a
126
+ Terraform module for AWS (S3, RDS, Secrets Manager, IAM role, Helm release).
127
+ - **Forecasting** (`Forecaster`):
128
+ - multiple series and multi-step horizons;
129
+ - rolling-origin backtests against seasonal-naive;
130
+ - conformal intervals and hierarchical reconciliation.
131
+ - **Causal inference** (`estimate_effect`): doubly-robust AIPW with overlap and balance
132
+ diagnostics. **Uplift:** `UpliftModel` and `qini`.
133
+ - **Anomaly detection** (`AutoAnomaly`): an ensemble of three detectors with plain-language
134
+ explanations.
135
+ - **AutoML:**
136
+ - conformal `predict_interval` and `predict_set`;
137
+ - `search="optuna"`, `imbalance="auto"`, `n_jobs`;
138
+ - `predict_with_proba` and `set_inference_threads`.
139
+ - **Serving:**
140
+ - Prometheus `/metrics` and OpenTelemetry tracing;
141
+ - canary and shadow rollouts, and `app_from_registry`;
142
+ - SageMaker, Vertex AI and Azure ML routes and Docker targets;
143
+ - an ASGI app and the `container` entry point;
144
+ - single-row latency down from about 500 ms to about 50 ms.
145
+ - **Security:**
146
+ - Ed25519 model signing, verified before unpickling whenever `AC_MODEL_PUBLIC_KEY` is set;
147
+ - `skops` export;
148
+ - signing on registration with `register(..., sign_with=...)`.
149
+ - **Monitoring:**
150
+ - PagerDuty and Opsgenie alert handlers;
151
+ - `join_labels` and `Monitor.check_labels` for labels that arrive later;
152
+ - `sql_sink` and `Monitor.on_check`;
153
+ - `SQLSource.insert` and `execute`.
154
+ - **Scale:**
155
+ - `transform_file` for files of any size;
156
+ - Spark, Ray and Dask scoring;
157
+ - `to_sql` SQL pushdown of fitted cleaning pipelines.
158
+ - **Integrations:** Airflow operators, Prefect tasks and flow, Dagster assets, Kafka stream
159
+ scoring, and plain `integrations.tasks`.
160
+ - **MLflow** (`integrations.mlflow`, `[mlflow]` extra):
161
+ - batched, idempotent export of runs and whole trackers;
162
+ - AutoML models logged as pyfunc models with a signature, so they serve with `mlflow models serve`;
163
+ - `export_registry` (stages become aliases) and `load_automl`.
164
+ - **DVC** (`integrations.dvc`, `[dvc]` extra): `DVCRepo` (init, remotes, track, push and pull,
165
+ read any revision, checkout, `log_to_run`) and `export_repository` for `DataRepository`
166
+ history.
167
+ - **Feast** (`integrations.feast`, `[feast]` extra): `export_to_feast` writes, applies and
168
+ materializes a Feast repository.
169
+ - **Data splitting** (`split_data`, `cv_folds`, `check_split`): stratified, grouped and time-ordered
170
+ splits with gaps; every split is leak-checked (shared groups, time overlap, duplicate rows, unseen classes).
171
+ - **Model security:** `robustness_report` (noise stability, a minimal adversarial attack,
172
+ feature sensitivity, extreme inputs) and `poisoning_report` (confident-learning label errors,
173
+ conflicting duplicates, backdoor triggers, outliers).
174
+ - **Environments** (`Environments`, `environments.yml`): isolated dev / QA / prod registries;
175
+ `release` enforces order, source stage, metric gates, four-eyes approvals and signatures, and
176
+ records provenance on both sides. The registry and Hub gain a `qa` stage.
177
+ - **Model CI/CD:** the CLI commands `train`, `gate`, `register`, `promote`, `release`, `latest`,
178
+ `deploy-sagemaker` and `cicd init` (GitHub Actions / GitLab CI pipelines, checked with actionlint).
179
+ - **Secrets** (`secretstore`): `vault://` (token, AppRole, Kubernetes; KV v1 and v2), `aws-sm://`,
180
+ `env://` and `file://` references, resolved wherever credentials are read.
181
+ - **Open Policy Agent for the Hub** (`hub.policy`): deny-overrides access rules, promotion rules
182
+ and fail-closed behaviour, plus Rego policies with unit tests in `deploy/opa`.
183
+ - **Serving:**
184
+ - `POST /feedback` with prediction IDs, live-accuracy metrics and `join_feedback`;
185
+ - hot reload (`watch=`, `POST /admin/reload`, registry-following);
186
+ - live drift, estimated-performance and alert gauges (`monitor=`).
187
+ - **Monitoring before labels:** `PerformanceEstimator` (CBPE, direct loss estimation),
188
+ `domain_classifier_drift`, and `estimated_performance` alerts in `Monitor.check`.
189
+ - **Grafana + Prometheus** (`deploy/monitoring`): a dashboard, 14 alert rules with `promtool`
190
+ unit tests, and a compose stack.
191
+ - **Lineage** (`lineage`): OpenLineage run events with schema, quality and version facets; HTTP
192
+ and file transports; `emit_tracked_run`, `emit_dataset_version`, `LineageGraph`.
193
+ - **Catalogs and orchestration:** AWS Glue (catalog tables, partitions, reads, job runs) and Azure
194
+ Data Factory pipeline runs.
195
+ - **AWS deployment** (`deploy.aws`): ECR push and blue/green SageMaker endpoints (serverless or
196
+ instances, autoscaling, data capture, rollback alarms); Terraform for SageMaker and ECS Fargate.
197
+ - **Evidently** (`integrations.evidently`): reports plus agreement with the built-in drift report.
198
+ - **Tests against real infrastructure:** Prometheus, Grafana, Vault, OPA and Marquez in Docker
199
+ (`AC_DOCKER_TESTS=1`); moto for AWS; dedicated CI jobs.
200
+ - These three are tested against the real tools (MLflow server and model server, git + DVC
201
+ with a remote, Feast), including in a dedicated CI job.
202
+ - **Image and audio data** (`media`):
203
+ - quality audits (corrupt files, exact and near duplicates, blur, exposure, clipping, silence,
204
+ mixed formats);
205
+ - preprocessors (EXIF-aware image resize and normalise; audio resample, trim and level);
206
+ - augmentation;
207
+ - features for AutoML (colour and edge-orientation histograms; MFCC and spectral features).
208
+ - **Statistical testing** (`stats`):
209
+ - automatic test selection with assumption checks, effect sizes and confidence intervals;
210
+ - chi-square and Fisher's exact test, correlations;
211
+ - A/B tests, power analysis, multiple-testing correction, `test_all`.
212
+ - **Data version control** (`DataRepository`): content-addressed commits, tags, row-level
213
+ diffs, checksum-verified checkout, links to experiment runs, garbage collection.
214
+ - **Feature store** (`FeatureStore`, `Entity`, `FeatureView`): point-in-time-correct
215
+ historical joins with TTL, an SQLite online store, `materialize`, schema guards.
216
+ - **Documentation site** (MkDocs Material with an API reference), plus `SECURITY.md`,
217
+ `CONTRIBUTING.md` and a versioning policy.
218
+
219
+ ### Removed
220
+ - `automatedcleaning/cleaning_update.py`: dead code that nothing imported.
221
+
222
+ ## [1.14.0] — 2026-11-08
223
+ - **Ingestion & connectors:**
224
+ - files, globs, HTTP, and S3 / GCS / Azure;
225
+ - any SQL database (bound parameters);
226
+ - Snowflake and BigQuery;
227
+ - REST APIs with five pagination styles;
228
+ - reusable schema inference and incremental loads.
229
+ - **Experiment tracking & model registry:** runs, dataset fingerprints, dev → staging →
230
+ production promotion with metric gates, audit history and rollback.
231
+ - **Deployment:**
232
+ - batch scoring and a REST API (API keys, prediction log);
233
+ - bundle, pickle and ONNX export;
234
+ - Docker build contexts and a CLI.
235
+ - **Monitoring:** data and prediction drift (PSI), accuracy decay, alerts and webhooks, and
236
+ champion-vs-challenger retraining.
237
+ - **Insights & reporting:** significance-tested findings, executive reports, and optional
238
+ narratives written by Claude.
239
+
240
+ ## [1.13.0] — 2026-11-06
241
+ - **Feature engineering:**
242
+ - dates, out-of-fold target encoding, interactions, aggregations, lags, text embeddings;
243
+ - feature selection and leakage detection.
244
+ - **AutoML:** task detection, time-budgeted model search, an ensemble, and AutoCluster.
245
+ - **Evaluation & explainability:** metrics with confidence intervals, built-in SHAP, error
246
+ analysis, fairness checks and model cards.
247
+
248
+ ## [1.12.0] — 2026-11-06
249
+ - Profiling & data quality: automated EDA, the 0–100 quality score, YAML data contracts, schema
250
+ diffs and quality gates.
251
+
252
+ ## [1.11.0] — 2026-11-05
253
+ - Shared `Step` / `Pipeline` core and the stateful cleaning pipeline (JSON save/load, audit
254
+ reports, schema checks).
255
+
256
+ ## [1.10.0] and earlier — 2026-07
257
+ - 1.10.0: 8 more data-quality checks in the dashboard.
258
+ - 1.9.0: column pickers in the dashboard.
259
+ - 1.8.x: the dashboard data-quality panel.
260
+ - 1.7.0: the enterprise dashboard redesign.
261
+ - 1.6.x: 43 additional cleaning steps and a dependency cleanup.
262
+ - 1.5.x: the Rust compute backend.
@@ -98,7 +98,7 @@ dependencies = [
98
98
 
99
99
  [[package]]
100
100
  name = "mlpilot-rustcore"
101
- version = "2.1.0"
101
+ version = "2.2.0"
102
102
  dependencies = [
103
103
  "once_cell",
104
104
  "pyo3",
@@ -1,6 +1,6 @@
1
1
  [package]
2
2
  name = "mlpilot-rustcore"
3
- version = "2.1.0"
3
+ version = "2.2.0"
4
4
  edition = "2021"
5
5
  description = "Rust compute backend for the MLPilot Python library"
6
6
  license = "MIT"