Crate chronix_analytics 

Chronix Analytics — Real-Time Streaming Analytics

Streaming anomaly detection, continuous forecasting, materialized forecast views, alerting, feedback, and forecast accuracy tracking.

Architecture

CDC PointWritten events
      │
      ├──→ StreamingAnomalyEngine (per-series anomaly detection)
      │          └──→ AlertEngine (configurable alert actions)
      │
      ├──→ ContinuousForecastEngine (incremental model updates)
      │          └──→ ForecastCache (materialized forecast views)
      │
      └──→ AnomalyPrecisionTracker (user feedback)
           ForecastAccuracyTracker (retroactive accuracy metrics)

Example

use chronix_analytics::{
    StreamingAnomalyEngine, StreamingAnomalyConfig,
    ContinuousForecastEngine, ContinuousForecastConfig,
};
use chronix_analytics::anomaly::DetectorType;
use chronix_analytics::forecast::ModelType;

// Set up streaming anomaly detection
let anomaly_engine = StreamingAnomalyEngine::new();
let config = StreamingAnomalyConfig::new("cpu", DetectorType::ZScore, 3.0);
anomaly_engine.enable(config).unwrap();

// Set up continuous forecasting
let forecast_engine = ContinuousForecastEngine::new();
let config = ContinuousForecastConfig::new("cpu", ModelType::Ses);
forecast_engine.enable(config).unwrap();

Re-exports

pub use error::AnalyticsError;

pub use registry::DetectorFactory;

pub use registry::ModelFactory;

pub use registry::PluginError;

pub use registry::PluginInfo;

pub use registry::PluginRegistry;

pub use registry::global_registry;

Modules

anomaly

chronix-anomaly

compute

Chronix Compute Engine

error

Error types for the analytics system.

forecast

chronix-forecast

lifecycle

Chronix Model Lifecycle Management

multivariate

chronix-multivariate

preprocess

Chronix Preprocessing Pipeline

registry

Thread-safe plugin registry for custom forecast models and anomaly detectors.

Structs

AccuracyMetrics

Accuracy metrics for a forecast model.

AlertConfig

Configuration for an alert.

AlertEngine

Alert engine — evaluates anomaly scores against configured alerts.

AnomalyFeedback

A feedback entry for an anomaly.

AnomalyPrecisionTracker

Precision tracker — computes rolling precision from feedback.

ContinuousForecastConfig

Configuration for continuous forecasting on a measurement.

ContinuousForecastEngine

Continuous forecast engine — re-fits models as new data arrives.

FiredAlert

A fired alert record.

ForecastAccuracyTracker

Forecast accuracy tracker — stores and queries accuracy metrics.

ForecastCache

Cache of materialized forecasts.

MaterializedForecast

A materialized (cached) forecast for a series.

PrecisionStats

Precision statistics.

ScoredAnomaly

A scored anomaly result from streaming detection.

StreamingAnomalyConfig

Configuration for streaming anomaly detection on a measurement.

StreamingAnomalyEngine

Real-time streaming anomaly detection engine.

Enums

AlertAction

Action to take when an alert fires.

FeedbackLabel

Feedback label for an anomaly detection.

ForecastUpdate

Update result from processing a point.

Functions

compute_accuracy

Compute accuracy metrics from forecasted vs actual values.

compute_accuracy_with_mase

Compute accuracy metrics including MASE using in-sample training data.

mase

Compute the Mean Absolute Scaled Error (MASE).