bigframes: v2.50.0

PyPI History

2.50.0 (2026-10-05)

Features

• add DataFrame.bigquery.ai.evaluate accessor (df96c47)
• add bigquery.ml.arima_coefficients function. (e6076aa)
• add bigquery.ml.arima_evaluate function. (2849013)
• add bigquery.ml.explain_forecast function (305cb55)
• add bigquery.ml.forecast function (791e40a)
• add bigquery.ml.recommend function (5223eb4)
• implement ai.predict() (19bf773)
• support bigframes.bigquery.ai.evaluate (AI.EVALUATE) (5aca1ed)
• support fill_value parameter on DataFrame.unstack() and Series.unstack() (6b49556)
• support indicator parameter on DataFrame.merge() and bpd.merge() (a1b243c)

Bug Fixes

• avoid installing prerelease dependencies in regular system tests (402f180)
• Make SQLGlot properly handle JSON fields backed by strings (3e333af)

Documentation

• add a requirements-docs.txt to fix the readthedocs build (df6405d)
• add Open in Colab Enterprise badge to remaining notebooks (5ae4375)
• add series AI accessor to the doc site (034ea50)
• deploy to readthedocs as an "unoffical" mirror (251c60e)
• enhance repository introduction page in README.rst (2ba6788)
• expand landing page with sample notebooks (52469d0)
• fix notebook header links and replace tree with blob (75fce7d)
• migrate BigQuery DataFrames samples from python-docs-samples (f617771)
• replace broken Open in Vertex AI Workbench links with Colab Enterprise (bd138c5)
• update imported ONNX and TensorFlow model snippets to use the bigframes.bigquery.ml module and partial ordering. (707c097)
• update repo location to google/bigframes (63defc9)