Skip to main content

Introduction to BigQuery ML (BQML)

What is BQML?​

BigQuery Machine Learning (BQML) is a feature inside Google Cloud's BigQuery that lets you build and run machine learning models using standard SQL queries.

Why use BQML?​

  • No need to move data: You can train models directly where your data lives.
  • Easy to learn: If you know SQL, you can do machine learning! No need to learn complex programming languages like Python or R.
  • Fast and scalable: It uses BigQuery's powerful engine to process massive datasets quickly.

The 4 Steps of BQML​

Building a machine learning model usually follows these 4 simple steps:

  1. Prepare Data: Clean and format your data into a table with features (behavior) and a label (what you want to predict).
  2. Create Model: Use CREATE MODEL to train your model using your prepared data.
  3. Evaluate Model: Use ML.EVALUATE to check how accurate your model is.
  4. Make Predictions: Use ML.PREDICT to guess future outcomes on new data.