Skip to content
Kitbay
Submit

bigquery-bigframes

Apache-2.0Updated on Nov 11, 202623,842GitHub stars

A Google skill that steers an agent when it writes Python against BigQuery with the BigQuery DataFrames library, BigFrames.

What the skill teaches

A Google skill that steers an agent when it writes Python against BigQuery with the BigQuery DataFrames library, BigFrames. It sets working rules: switch on partial ordering mode, preview with peek rather than head, keep work in BigQuery instead of pulling data into pandas, use DataFrame methods and accessors in place of raw SQL or lambdas, and check schemas of intermediate results. For machine learning it directs the agent to the bigframes.bigquery.ml package, with reference notes on linear and logistic regression. It is meant as the default for Python data tasks on BigQuery, not for SQL-first work. It lives in Google's skills repository.

Capabilities

  • Generate Python code with BigQuery DataFrames
  • Keep processing inside BigQuery instead of local memory
  • Train regression models with the BigFrames ML package
  • Plot results from BigFrames objects