Nota
L-aċċess għal din il-paġna jeħtieġ l-awtorizzazzjoni. Tista’ tipprova tidħol jew tibdel id-direttorji.
L-aċċess għal din il-paġna jeħtieġ l-awtorizzazzjoni. Tista’ tipprova tibdel id-direttorji.
This article provides an overview of the specifications and capabilities of the anomaly detection models available in Fabric Real-Time Intelligence. These models are designed to automatically identify unusual patterns and outliers in your data streams.
Important
This feature is in preview.
Supported models
| Model Name | Description | Package |
|---|---|---|
| Signal Watcher | Analyzes the underlying signal to detect unusual behaviors, from subtle shifts to sharp spikes. | TSB-AD - Based on SR algorithm |
| Signal Watcher (Seasonal) | Detects a wide range of unusual behaviors, from subtle shifts to sharp spikes, by analyzing the underlying signal augmented with seasonality. | TSB-AD - Based on SR algorithm |
| Signal Watcher (Enhanced Seasonal) | Detects a wide range of unusual behaviors, from subtle shifts to sharp spikes, by analyzing the underlying signal, augmented with complex seasonality. | TSB-AD - Based on SR algorithm |
| Histogram Sentinel | Identifies anomalies based on data distribution patterns, offering fast and scalable performance for large datasets. | TSB-AD - Based on HBOS algorithm |
| Pattern Proximity | Uses k-nearest neighbors to detect anomalies based on the proximity of data points in the feature space. Ideal for local pattern shifts. | TSB-AD - Based on KNN algorithm |
| Core Pattern Finder | Reduces complex data to its most essential patterns, making it easier to detect subtle and hidden anomalies. | TSB-AD - Based on PCA algorithm |
| Change Spike Detector | Spots sharp, local changes by comparing how values evolve over time. | MS Developed |
| Rolling Change Tracker | Tracks moving trends to identify gradual shifts in data patterns. | MS Developed |
| Outlier Radar | Highlights data points that deviate significantly from the average, useful for spotting large and sudden outliers. | MS Developed |
| Robust Outlier Radar | Similar to Outlier Radar, this model uses the median for a more robust analysis of skewed data. It focuses on significant deviations while ignoring natural fluctuations. This makes it stable in noisy environments. | MS Developed |
| Robust Outlier Radar (Seasonal) | Handles complex data distributions and incorporates seasonal awareness, making it ideal for recurring patterns. | MS Developed |
| Deviation Pulse | Monitors signals for significant deviations, optimized for detecting standout events. | MS Developed |