Skip to content

Data Preview

Data preview (admin) lets you inspect raw and processed series data and dry-run detectors on real loaded data before you rely on production schedules.

Open it from:

  • Left nav → Data preview (/datapreview)
  • A data view’s top-right Data preview (search) button → /dataviews/{id}/preview

The page starts empty until you select a data view.

Data Preview

Workflow

  1. Select a data view from the dropdown (domain / name).
  2. Optionally narrow categories (checkbox per dimension, then pick a value).
  3. Optionally edit Detectors YAML and Special dates YAML.
  4. Click Display analysis.

Data Preview

Data view selection

Choose from all available data views. Opening preview from a data view detail pre-selects that view and loads its detector / special-date settings.

Data Preview

Category filters

Below the data view selector, checkboxes appear for each category defined on the data view. Enable a category and pick a member value to restrict the series (for example a product type in a region on a given sales channel). Which values appear depends on the data already loaded for that data view.

With no category checkboxes enabled, analysis uses the default / unfiltered series path returned by the analysis engine for that data view.

Data Preview

Detectors and special dates

After a data view is selected, two YAML editors appear:

Editor Purpose
Detectors Detector YAML for the preview run. The page loads the data view’s first detector by default. Use Add detector to insert a catalog sample (in preview this replaces the first detector entry so you can fine-tune one detector at a time).
Special dates Optional special-date YAML for the run. Use Sample for a template. Rules apply when a detector has useSpecialDates: true.

Edits here are for the preview session only — they are not written back to the data view until you copy them into the data view Analysis / Special Dates tabs and save.

Data Preview

Use this UI to fine-tune detector parameters on real series and reduce false positives or false negatives before enabling cron/API schedules.

Results

After Display analysis, the right panel shows:

Section When shown Contents
Raw data Series found Chart of loaded points, plus Average value and Gap amount [%]
Processed data Processed series non-empty Chart after gap-fill / preprocessing used by detectors
No data for this combination Empty series Selected category filters match no records
Detectors skipped One or more detectors skipped Detector id + reason (for example gap amount too high, average below minValue, or above maxValue) — see Detectors shared series filters
Anomalies Detectors were configured for the run Table: date, detector, duration, change %, change, share %, recurrence; direction marker (positive / negative). Hidden for chart-only runs (empty detectors YAML)

Detectors are optional. With no detectors configured, the page still renders series charts when data exists.

Prerequisites and tips

  • The data view should be deployed and have data loaded; otherwise there may be nothing to chart.
  • Preview is admin-only.
  • Category member lists come from loaded fact categories (GetTopNFactCategories) — load data first if the lists look empty.
  • Prefer validating one detector at a time in preview, then copy the tuned params into the data view Analysis YAML (where you can keep multiple detectors).
  • Pair special dates with useSpecialDates: true to check calendar suppression — see Special settings.