Docs/Monitoring

✦

In this section

Part of Celiq's semantic layer platform. Connect your warehouse, model your data once, query it everywhere.

ML & Predictions

What it is: Machine learning models running on your warehouse data — forecasting future values and detecting statistical anomalies. Why use it: Know what will happen, not just what did. Get genuine anomalies without alert fatigue from rule-based thresholds. Where to find it: Ask Orion any forecast or anomaly question. Forecast tiles are also available in Mosaic → Add tile → Forecast tile. Who can use it: All roles.

What Celiq ML can do

Capability	How to use it
Forecast any metric	Ask Orion: "What will revenue be next month?"
Detect statistical anomalies	Ask Orion: "Are there any anomalies in orders?"
Forecast tile in Mosaic	Add tile → 📈 Forecast tile, configure metric + date field
Orion Pulse upgrade	Pulse alerts now use seasonal decomposition, not simple percentage thresholds

How it works

Celiq ML is a separate microservice (powered by Meta Prophet) that Orion calls when you ask a predictive question.

Your question → Orion detects ML intent
             → Pulls historical data from your warehouse
             → Sends to ML microservice
             → Prophet fits a seasonal model
             → Returns forecast / anomalies
             → Orion explains results in plain English

The ML service accounts for weekly and yearly seasonality, so anomaly detection won't flag normal Monday/weekend patterns as anomalies.

Topics

Forecasting — predict any metric forward in time
Anomaly detection — find genuinely unusual data points