sequence_detect plugin

Detects sequence occurrences based on provided predicates. The plugin is invoked with the evaluate operator.

Syntax

T | evaluate sequence_detect (TimelineColumn, MaxSequenceStepWindow, MaxSequenceSpan, Expr1, Expr2, …, Dim1, Dim2, …)

Parameters

Returns

Returns a single table where each row in the table represents a single sequence occurrence:

• Dim1, Dim2, …: dimension columns that were used to correlate sequences.
• Expr1TimelineColumn, Expr2TimelineColumn, …: Columns with time values, representing the timeline of each sequence step.
• Duration: the overall sequence time window

Examples

The following query looks at the table T to search for relevant data from a specified time period.

T | evaluate sequence_detect(datetime_column, 10m, 1h, e1 = (Col1 == 'Val'), e2 = (Col2 == 'Val2'), Dim1, Dim2)

Exploring Storm Events

The following query looks on the table StormEvents (weather statistics for 2007) and shows cases where sequence of ‘Excessive Heat’ was followed by ‘Wildfire’ within 5 days.

StormEvents
| evaluate sequence_detect(
               StartTime,
               5d,  // step max-time
               5d,  // sequence max-time
               heat=(EventType == "Excessive Heat"), 
               wildfire=(EventType == 'Wildfire'), 
               State
           )

Output