Metrics

Metrics are rarely steady. If you are monitoring a number of orders or your global revenue, those numbers will probably vary according to different seasonality. Your shops might be closed on Sunday, causing a scheduled drop in the figures or you may have an raise in sales during summer because you are a sunglasses selling company. You might also not be sure what to expect.

For all those reasons, monitoring metrics with machine learning models make sense. With Sifflet, you can:

• Monitor a statistical transformation of a numerical field with dynamic thresholds ->
• Write a SQL query to monitor a custom metric that would not be stored in a table
• Compare two different metrics to make sure they are aligned