GCP BQML

BQML enables data scientists to create and train ML models directly in BigQuery using standard SQL queries. BQML improves the ML model development speed by eliminating the need to move data and directly using BigQuery datasets as training and testing datasets. BQML-trained models can be exported directly to Vertex AI (to be discussed in later chapters) or other cloud serving layers.

BQML can be accessed and used in the following ways:

• The GCP console via a web browser
• The bq command-line tool via Google Cloud Shell or a VM shell
• The BigQuery REST API
• External tools such as Jupyter Notebook

As we discussed in Chapter 3, Preparing for ML Development, and Chapter 4, ML Model Developing and Deploying, the ML process includes data preparation, model creation and training, model validation/evaluation, and model deployment/prediction. Let’s go over this process with BQML.

The first step is data preparation. With BQML, you can prepare the training...