RemoveAboveValue

Description 

 Removes data samples from each input time-series that are larger than a specified threshold value. 

Input 

 Time Series: Multiple (Single or multiple time-series) 

Parameters 

 Threshold 

  Mandatory, numeric  

Output 

 Transformed Time Series: Multiple 

Available in 

 Alerts, Composites, Dashboards  

The RemoveBelowValue function is useful when the underlying data is known to contain data spikes which are irrelevant or invalid, and should be removed. 

Usage example

An Ad-tech company monitors the cost of several advertising campaigns. Occasionally, a bug is introduced in the data collection process, which results in a huge outlier: 

RemoveAboveValue_0.png

This spike, however, does not reflect a valid value, and should be removed. By applying the RemoveAboveValue function (specifying a threshold of 1000) the spike can be removed from the time series, and the normal pattern is clearly evident: 

removeAboveValue2.png

removeAboveValue.png

Note: the RemoveBelowValue function can be used when the invalid samples are not spikes, but drops.