Thresholds
Datachecks offers two types of thresholds to monitor your data attributes effectively: Auto Thresholds and Constant Thresholds.
Auto Thresholds
Auto Thresholds automatically generate thresholds for every data attribute you track using a proprietary machine-learning engine, providing meaningful and actionable alerts without any manual effort.
How Auto Thresholds Work
Auto Thresholds learn from your historical data, taking into account factors like seasonality and trends, allowing them to adapt to natural changes in your data over time. The underlying structure of the data series is analyzed through preliminary statistical tests, and the expected range is fine-tuned based on user-defined settings.
Auto Thresholds are periodically updated to reflect the latest changes in your data, incorporating new data into the training history. They are also refreshed on a set schedule to ensure they remain accurate and responsive to evolving data patterns.
Constant Thresholds
Constant Thresholds are predefined limits that you set for specific metrics or data attributes. This approach allows you to establish fixed values that trigger alerts whenever the monitored data exceeds or falls below these thresholds.
Constant Thresholds require manual configuration, enabling you to tailor alerts to your specific business needs and operational criteria. You define the threshold value based on your understanding of the data and its expected behavior.
For example, you might set a threshold to alert you when the number of rows in a dataset exceeds 30. This approach is straightforward and effective for scenarios where you anticipate consistent behavior in the data or want to enforce strict compliance with certain standards.
While Constant Thresholds provide clarity and control, they may not adapt to fluctuations in data that occur over time.
Updated 3 days ago