BQML

We first introduced BigQuery in Chapter 3, and we’ve used it in various chapters of this book for its data management and processing functionality. However, given the close relationship between large-scale data processing and ML, Google Cloud has built ML functionality directly into BigQuery, as well as native integrations with Google Cloud Vertex AI. This functionality is referred to as BQML, and this section explores this service in detail.

BQML enables us to create and execute ML models using standard SQL queries in BigQuery. Considering that many companies already store large amounts of data in BigQuery, BQML makes it easy for data scientists in those companies to train models on large datasets and make predictions directly in the database system, without needing to move the data around between different storage systems.

It supports a variety of ML algorithms, including linear regression, logistic regression, k-means clustering, and deep neural networks, as well...