Google BigQuery is a serverless, highly scalable and cost-effective cloud data warehouse designed for business agility. Handling both real-time and batched data, many companies are using BigQuery for their machine learning efforts.

Google Big Query: Example Use Cases

BigQuery enables machine learning capabilities to build and scale inside the Google Cloud Platform. There are a lot of different ways to use BigQuery. For example, you may be an organization with many data scientists. An approach for you might include using your model frameworks and code to leverage BigQuery.

With TensorFlow models, you can train those models on an AI platform, save them to Google Cloud Storage and then upload those models to BigQuery. Your model is then stored in the data warehouse and served in real-time. As a result, you can leverage standard ML tooling and libraries while taking advantage of the BigQuery platform to serve your model.

Perhaps you understand data, but don’t possess a background in data science. You may want to use the tools and methods in BigQuery such as TensorFlow, without experience in opensource frameworks. BigQuery supports data warehouse machine learning. You’ll have support for classification, linear regression, and matrix factorization. Plus, you can import your TensorFlow models.

Google BigQuery & AutoML Tables

Are you an analyst who isn’t all that technical? BigQuery can help.

You can quickly run and test out different

models  using AutoML Tables. This tool has a point-and-click UI, allowing you to point AutoML tables to a BigQuery table. Simply tell AutoML Tables which field you want to predict and give it a runtime budget to receive your final model.

AutoML is fully automated, as long as your data is set up in BigQuery. Feature engineering and hyperparameter tuning are handled behind the scenes. AutoML will deliver information such as nullability, invalid values and correlation with a target. Right out of the gate, it takes a lot of the legwork out of the typical ML workflow even before you train your model.

Moving Into Production

AutoML makes it easy to go back and retrain your models. For example, maybe you want to retrain your model to remove unimportant features or move forward with your existing model. It’s easy to:

  • Batch predictions if you want to point your model at a new dataset in BigQuery for predictions
  • Generate online predictions to get predictions in real-time
  • Export your model to serve it locally on a device

Enhance Your Machine Learning Capabilities With Google Cloud Platform

Google Cloud Platform and BigQuery make it simple to gain unique insights and identify emerging patterns in your business. To learn more about machine learning and how you can use BigQuery to enhance your capabilities, give us a call at 612-430-6316 or send us a message.