Sustained Shift

YAML type: sustain | Category: Adaptive

Detects run-length deviations that stay above/below baseline for multiple days.

Description

Sustain requires the series to remain away from a rolling baseline for at least minDurationDays. Cooldown reduces re-firing.

How it works

Sustained Shift illustration

Sustain tracks how long values stay above or below a rolling baseline by at least deviationPercent. Only when that run reaches minDurationDays does it emit an anomaly; coolDownDays then reduces immediate re-firing.

Use it when the business cares about persistent shifts, not one-day blips. Baseline method (median or mean) changes how robust the reference level is.

Setup

Tune deviationPercent, baselineDays, and coolDownDays to match how long a real shift lasts.

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

Settings

Parameter	Default	Description
`baselineDays`	`30`	Rolling baseline window
`minDurationDays`	`4`	Minimum sustained duration
`deviationPercent`	`25`	Required % deviation from baseline
`mode`	`both`	Direction -- see Allowed values
`coolDownDays`	`7`	Suppress re-fire after an alert
`baselineMethod`	`median`	Baseline statistic -- see Allowed values

Allowed values

`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

`baselineMethod`

Value	Meaning
`median` (default)	Rolling baseline is the median of the prior `baselineDays`
`mean`	Rolling baseline is the mean of the prior `baselineDays`

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: sustain
      params:
        baselineDays: 30
        minDurationDays: 4
        deviationPercent: 25.0
        mode: both
        coolDownDays: 7
        baselineMethod: median

Using the detector

Prefer over Spike when the business cares about persistent shifts.

Use cases

Prolonged conversion drops
Multi-day outages reflected in KPIs
Post-change performance regressions