# Anomaly Analytics Overview

Slingshot’s Anomaly Analytics view uses Agatha AI, our AI Anomaly Detection and Characterization Engine, to identify unusual or unexpected satellite behaviors and characteristics using high-quality, multi-source space domain data.

April 7, 2026

## What The Analytics Provides

Out of Family Analytics:

- Continuously monitors satellite behavior
- Identifies statistically significant deviations
- Ranks spacecraft by “Interest Factor”
- Surfaces high-priority objects for investigation
- Provides explainable metrics to support operational decisions

## Operational Cadence

AI Anomaly Detection runs on a scheduled cadence:

- **Frequency:** Daily
- **Lookback Window:**
  - LEO: 45 days
  - GEO: 14 days
- **Coverage:** Focused on satellites launched after 2010

## Understanding the Interest Factor

The **Interest Factor** is a relative measure of how unusual a satellite’s recent behavior is compared to its baseline.

- Values range from 0 to 1
- Higher values indicate greater deviation
- Used to rank satellites for investigation

Important:

The Interest Factor is comparative — it highlights objects that stand out relative to peers and historical behavior.

## Satellite Interest Factors Chart

The primary visualization displays satellites plotted by NORAD ID and Interest Factor score.

- Each point represents a satellite
- Color coding reflects relative deviation
- Date of run is displayed above the chart

## Filtering Options

You can refine results using:

- **Orbit Regime Toggle:** LEO or GEO
- **Date Selector:** Select analysis date
- **Filter Icon:** Apply additional filters

## Top Five Spacecraft of Interest

The system highlights the top-ranked spacecraft based on Interest Factor.

For each object, you’ll see:

- Name
- Country
- Orbit regime
- Launch year
- Key contributing factors (e.g., orbital motion, brightness variability)

## Investigation Panel

Selecting a spacecraft opens the **Investigation View**.

This view includes:

- Satellite metadata
- Country
- Orbit regime
- Launch year
- Interest Factor
- **Compare** – Compare with other spacecraft
- **View Metadata** – Access detailed satellite information
- **View in 3D** – View object in Portal’s Digital Space Twin.

## Investigation Metrics

Each spacecraft includes detailed expandable analytics panels in the **Investigation** tab.

At a high level, the Investigation metrics help users answer questions such as:

- What type of behavior is making this object notable?
- Is the anomaly related to orbital motion, brightness, proximity, or position-keeping?
- Does the object appear to be behaving differently from its peers or from its own recent history?

The metrics shown can vary depending on whether the selected object is in **LEO** or **GEO**.

### LEO metrics

For **LEO** objects, Investigation metrics are typically focused on behaviors such as:

- **Semi-Major Axis,** to understand changes in orbital altitude of the object
- **Close Approach**, to assess proximity to other objects
- **Apparent Magnitude**, to evaluate changes in observed brightness over time
- comparisons to **Cluster**, **Shell**, and **Ring** reference populations

These metrics are most useful for understanding whether a LEO object is behaving differently from similar objects in nearby orbital environments.

### GEO metrics

For **GEO** objects, Investigation metrics are typically focused on behaviors such as:

- **Longitude Drift Rate**, to understand east-west motion
- **Longitude**, to show position along the geostationary arc
- **Inclination**, to track changes in orbital plane
- **Close Approach**, where relevant, to assess proximity to nearby objects
- brightness-related metrics such as **Apparent Magnitude** when available

These metrics are most useful for understanding position-keeping, relocation, and other changes in geostationary behavior.

## Using Compare

The **comparison view** is important because our anomaly detection engine isn’t just about seeing a pattern — it is about seeing whether that pattern is unusual relative to the population. Comparison is what turns a chart from an interesting trend into a more meaningful behavioral signal.

Use this to:

- Identify relative behavior differences
- Validate anomaly severity
- Contextualize activity within a peer group

## Exporting Data

Each investigation metric includes export options as PNGs.

## Best Practices

For effective use:

- Start with the Top Five Spacecraft
- Investigate sustained Interest Factor spikes
- Cross-reference brightness with altitude changes
- Compare to other satellites in family
- Use metadata to contextualize behavior

## When to Escalate

Consider deeper analysis if:

- Interest Factor .01 (For Leo)
- Sustained altitude or drift changes
- Repeated close approaches
- Combined brightness and orbital anomalies

## Powered by Agatha AI

AI Anomaly Detection is powered by Slingshot’s Agatha AI framework, enabling multi-source fusion and built-in explainability.

## Next Steps

After identifying an anomaly, you can:

- Monitor the object in Operations
- Submit a tracking request
- Compare against similar satellites
- Export data for external reporting

## You’re Ready to Investigate 🚀

AI Anomaly Detection gives you advanced, explainable analytics to surface and investigate unusual satellite behavior with confidence.