Provide explainability for the model, to
clarify how particular features influenced a given prediction and also the
model overall.
Learn more about the components that comprize the model by using
model weights.
Because you can use many different kinds of models in BigQuery ML,
the functions available for each model vary. For more information about
supported SQL statements and functions for each model type, see the following
documents:
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Hard to understand","hardToUnderstand","thumb-down"],["Incorrect information or sample code","incorrectInformationOrSampleCode","thumb-down"],["Missing the information/samples I need","missingTheInformationSamplesINeed","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2025-09-04 UTC."],[[["\u003cp\u003eBigQuery ML enables the creation and operationalization of machine learning models using SQL over BigQuery data.\u003c/p\u003e\n"],["\u003cp\u003eModel development in BigQuery ML involves creating, preprocessing, tuning, evaluating, inferencing, and explaining models.\u003c/p\u003e\n"],["\u003cp\u003eBigQuery ML supports both automatic and manual feature preprocessing via functions and the \u003ccode\u003eTRANSFORM\u003c/code\u003e clause.\u003c/p\u003e\n"],["\u003cp\u003eHyperparameter tuning is used to refine the model to better fit the training data.\u003c/p\u003e\n"],["\u003cp\u003eThe available functions vary between each type of model, detailed in the end-to-end user journey for each model.\u003c/p\u003e\n"]]],[],null,["# Model creation\n==============\n\nBigQuery ML lets you build and operationalize machine learning (ML)\nmodels over data in BigQuery by using SQL.\n\nA typical model development workflow in BigQuery ML looks similar\nto the following:\n\n1. Create the model using the [`CREATE MODEL` statement](/bigquery/docs/reference/standard-sql/bigqueryml-syntax-create).\n2. Perform feature preprocessing. Some preprocessing happens [automatically](/bigquery/docs/reference/standard-sql/bigqueryml-auto-preprocessing), plus you can use [manual preprocessing functions](/bigquery/docs/reference/standard-sql/bigqueryml-preprocessing-functions) inside the [`TRANSFORM` clause](/bigquery/docs/reference/standard-sql/bigqueryml-syntax-create#transform) to do additional preprocessing.\n3. Refine the model by performing [hyperparameter tuning](/bigquery/docs/hp-tuning-overview) to fit the model to the training data.\n4. [Evaluate the model](/bigquery/docs/evaluate-overview) to assess how it might perform on data outside of the training set, and also to compare it to other models if appropriate.\n5. [Perform inference](/bigquery/docs/inference-overview) to analyze data by using the model.\n6. Provide [explainability](/bigquery/docs/xai-overview) for the model, to clarify how particular features influenced a given prediction and also the model overall.\n7. Learn more about the components that comprize the model by using [model weights](/bigquery/docs/weights-overview).\n\nBecause you can use many different kinds of models in BigQuery ML,\nthe functions available for each model vary. See the\n[End-to-end user journey for each model](/bigquery/docs/e2e-journey) to see\nthe specific functions available for each model."]]