SPLK-3002 practice question 14 of 15

What is the minimum number of entities a KPI must be split by in order to use Entity Cohesion anomaly detection?

Full set: 90 questions
Answer options
A: 3 Correct answer
B: 4
C: 5
D: 2
Correct Answer:
A. 3
Explanation

The Entity Cohesion anomaly detection algorithm in ITSI is designed to identify entities that are behaving differently from their peers. It uses a clustering algorithm (DBSCAN) to group entities with similar behavior patterns. To form a statistically meaningful cluster and reliably identify an outlier, the algorithm requires a minimum of three entities for the KPI being analyzed. With fewer than three entities, it is not possible to determine which entity is the "anomaly" versus just normal variation between two distinct entities.

Why Incorrect

B: While the algorithm works with 4 entities, it is not the minimum required number. C: While the algorithm works with 5 entities, it is not the minimum required number. D: Two entities are insufficient for the clustering algorithm to reliably identify an anomalous entity from its peer group.

References

- Splunk ITSI Documentation, Analytics and Algorithms, "Apply anomaly detection to a KPI". This documentation explicitly states, "The Entity Cohesion algorithm requires that the KPI is split by at least 3 entities."

Scroll to Top