Peer Group

YAML type: peer | Category: Business

Compares a series to peers sharing groupBy dimensions while differing on varyBy.

Description

Peer uses precomputed peer median/MAD baselines. It runs only on rollup series where groupBy and varyBy are set and all other categories are null.

How it works

Peer Group illustration

Peer builds a cohort of series that share groupBy dimensions and differ on varyBy, using precomputed peer median/MAD baselines from load time. Each eligible series is scored with robust_z or ratio against that peer set.

Detection runs only on rollup grains where groupBy and varyBy are set and all other categories are null. If Data Preview skips peer, check that grain rule first.

Setup

Define groupBy and varyBy category lists. Choose scoreMethod (robust_z or ratio) and thresholds.

Configure detectors on the data view Analysis tab or in YAML. Validate with Data Preview before enabling production schedules.

Settings

Parameter	Default	Description
`groupBy`	--	Shared peer dimensions
`varyBy`	--	Dimension that differs across peers
`minPeers`	`5`	Minimum peers required
`scoreMethod`	`robust_z`	Scoring method -- see Allowed values
`zThreshold`	`3`	For robust_z
`ratioLow` / `ratioHigh`	`0.6` / `1.6`	For ratio method
`mode`	`both`	Direction -- see Allowed values
`excludeSelf`	`true`	Exclude series from its peer set

Allowed values

`scoreMethod`

Value	Meaning
`robust_z` (default)	Compare to peer median using MAD; uses `zThreshold`
`ratio`	Compare to peer expected via ratio bounds `ratioLow` / `ratioHigh`

`mode`

Value	Meaning
`above`	Flag increases / values above the comparison (or high side of a band)
`below`	Flag decreases / values below the comparison (or low side of a band)
`both`	Flag either direction

Shared series filters

Most detectors also accept optional series filters in params:

Parameter	Purpose
`maxGapAmount`	Skip series when gap amount is too high (`null` = no limit). Catalog create-default is often `5`.
`minValue`	Skip series whose average value is below this (`null` = no limit). Catalog create-default is often `1000`.
`maxValue`	Skip series whose average value is above this (optional; omit/`null` = no limit)
`useSpecialDates`	When `true`, suppress anomalies that overlap special dates

When a filter skips a series, Data Preview shows the reason.

Configuration example

analysis:
  detectors:
    - type: peer
      params:
        groupBy: [region, channel]
        varyBy: [product]
        minPeers: 5
        scoreMethod: robust_z
        zThreshold: 3.0
        mode: both
        excludeSelf: true

Using the detector

If peer is skipped in Data Preview, check grain: other categories must be null on the series.

Use cases

Product vs peers in the same region/channel
Store performance vs peer stores
Detecting under/over-performers in a cohort

Specifics

Baselines are built at load time in a per-data-view peer baseline table.