Easily identify data irregularities with anomaly detection in Connected Sheets

You can now easily identify critical irregularities and outliers in your time-series data when using Connected Sheets to analyze BigQuery data sets from Google Sheets. Anomaly detection in Connected Sheets allows users to distinguish between expected trends and true outliers without requiring manual model training or complex SQL knowledge.

Powered by BigQuery ML and TimesFM, this capability delivers "zero-shot" analysis, meaning you can uncover actionable AI insights immediately—no need to configure or wait for a model to be trained on your data. Anomaly detection also builds upon our recently introduced forecasting feature to provide a more comprehensive suite of predictive tools right where you already work.

Key features include:

  • Easy, SQL-free configuration: A user-friendly side panel guides you through configuring your anomaly detection analysis without writing a single line of SQL.
  • Clear, built-in formatting: Results are cleanly rendered with new columns for is_anomaly (a boolean indicator) and lower_bound/upper_bound intervals to help you quickly sort, filter, and interpret the findings. 
  • Customizable thresholds: Take control of your analysis by setting a specific time period, anomaly probability threshold (defaulting to 0.95), and filtering input data
  • Automated refresh ability: Just like other Connected Sheets objects, your anomaly detection extracts can be scheduled to refresh automatically, ensuring your insights are always current.

Getting started

  • Admins: There is no admin control for this feature.
  • End users: Access to anomaly detection in Connected Sheets requires permissions to a BigQuery project with billing enabled.

Rollout pace

Availability

  • Available to all Google Workspace customers and users with personal Google accounts

Resources

This entry was posted in Uncategorized on by .