> ## Documentation Index
> Fetch the complete documentation index at: https://docs.dqlabs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

> What a Behavioral metric is in Prizm — a user-defined check that models expected data behavior over time and flags deviations from that model.

<script type="application/ld+json">
  {`{
            "@context": "https://schema.org",
            "@type": "TechArticle",
            "headline": "Behavioral Metrics Overview",
            "description": "What a Behavioral metric is in Prizm - a user-defined check that models expected data behavior over time and flags deviations from that model.",
            "url": "https://docs.dqlabs.ai/architecture/metrics/behavioural/overview",
            "publisher": {
              "@type": "Organization",
              "name": "DQLabs Inc",
              "logo": "https://media.brand.dev/332adc35-5bc4-4d2b-bf78-256aa4a5e414.svg"
            }
            }`}
</script>

## What is a Behavioral metric?

Behavioral metrics detect **abnormal patterns, trend breaks, and distribution shifts** in business or operational data over time. Where other metric types evaluate the current state of the data, behavioral metrics evaluate how the data is changing — relative to its own history, peer segments, or prior periods.

**Type:** User-defined — Behavioral metrics are created manually. Navigate to **Metric → Add** and select **Subcategory: Behavioral**.

Behavioral metrics answer the question: *Is this data behaving consistently with how it has always behaved, or has something fundamentally changed?*

Prizm builds a statistical model of the metric's expected behavior from historical runs. On each new run, the observed value is compared against the model's expected range. Deviations that exceed the configured sensitivity threshold are flagged as anomalies. The model accounts for daily and weekly patterns, growth trends, and historical variance — adapting continuously as data patterns evolve.

Behavioral metrics require a minimum of **5 historical data points** before anomaly detection activates.

### Metrics

| Metric     | Category | Sub Category | Asset Level       | Description                                                                                                                                                                                                                                                             |
| ---------- | -------- | ------------ | ----------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Behavioral | Business | Behavioral   | Asset / Attribute | A metric that models expected data behavior over time and flags deviations from that model. Uses adaptive baselines, grouping dimensions, and configurable comparison contexts — including historical windows, peer group comparisons, and period-over-period analysis. |

<CardGroup cols={2}>
  <Card title="Setup" icon="sliders" href="/architecture/metrics/behavioural/setup">
    Step-by-step guide to creating and configuring a behavioral metric.
  </Card>

  <Card title="Use Cases" icon="lightbulb" href="/architecture/metrics/behavioural/usecases">
    Real-world examples — transaction volume trends, ML feature drift detection.
  </Card>
</CardGroup>