Google AutoML Tables Connected System

Overview

Google AutoML Tables Connected System plug-in provides a way to use Google's Table Analytics capabilities to make predictions on structured data and keep models up to date.

Key Features & Functionality

The plug-in provides a wide range of functionality including:

  • Creating a dataset
  • Importing data into a dataset (supports import from Google Storage or Google Big Data)
  • Get process status
  • Get table metadata
  • Setting a target column for prediction
  • Updating a column configuration
  • Creating and training a model from a dataset
  • Publishing and unpublishing a model
  • Making a prediction
  • Getting and listing evaluation metrics for a model
  • Listing datasets and models
  • Deleting datasets and models
  • Getting target column id
  • Getting and setting weight columns

NOTE: Training and prediction cost money and require a GCP account with a service account enabled for billing in order to use these services. Training data must be loaded into Google Storage that MUST have US-Central1 as its region.

Anonymous