azure-ai-anomalydetector-java

You can detect anomalies in time-series data within Java by installing the azure-ai-anomalydetector SDK, configuring your endpoint, and authenticating with AzureKeyCredential or DefaultAzureCredential to run univariate or multivariate detection.

Real-time anomaly detection for IoT sensor networks is supported through last-point analytics, enabling Java applications to identify sudden metric deviations and trigger automated monitoring alerts on operational dashboards.

Univariate anomaly detection analyzes a single variable's metrics for deviations, while multivariate detection evaluates correlations across multiple time-series data streams simultaneously to identify broader anomalies in Java applications.

Authentication for the Azure AI Anomaly Detector SDK in Java requires configuring your service endpoint alongside credentials, utilizing either AzureKeyCredential for direct key access or DefaultAzureCredential for identity-based authentication.

Batch inference is fully supported for analyzing historical time-series data, allowing Java applications to train detection models and run comprehensive anomaly evaluations across large datasets for operational analytics.

The Azure AI Anomaly Detector requires proper endpoint configuration and valid Azure credentials, and is scoped to univariate and multivariate time-series data, meaning it cannot detect anomalies in unstructured or non-temporal data formats.