Documentation

Everything you need to get data in, understand model output and run investigations.

Quick start

Create a workspace, load the labelled sample dataset and run detection in under five minutes.

CSV import format

Columns: station_code, parameter, value, unit, measured_at (ISO 8601), optional calibration_date. Each row is validated and rejected rows are reported line by line.

Ingestion API

POST /api/public/ingest with header x-api-key and a JSON body { readings: [...] } using the same fields as CSV. Create keys in Organization settings.

Anomaly detection

Robust z ≥ 3.5 vs a 48-point trailing median and Isolation Forest score ≥ 0.6. Requires at least 30 points per series.

Review ranges

Admins set min/max per parameter with a recorded basis. Breaches are stored separately from model flags.

Investigations

Open a case from any alert, assign reviewers, link samples and evidence, then close with a recorded decision.

Watch the model score a stream in real time.

Isolation Forest + robust baselines, running in your browser.

Open live demo Create workspace