You are here:

Incident Cluster

Incident Cluster iCube enables you to create ad hoc reports that help you to categorize incidents by various topics or clusters using the ML technique of topic modeling. Identifying patterns and determining the root causes at a cluster level will help in effective incident management. Using this iCube, you can create interactive reports that focus on cluster health and provide better insights into incidents associated with specific clusters.

The Incident Cluster iCube comprises of the Incident Cluster and Cluster Description datasets

Incident Cluster dataset

Enables you to view information about incidents and the clusters that they belong to, along with the match probability for each incident. You can also view more details about incidents and analyze month-on-month statistics of incident volumes.

Attributes

Expand this link to view the Incident Cluster attributes:

Attributes	Description
#1 Cluster Match	Top cluster match for an Incident as determined by the Incident Clustering model
#1 Cluster Match Probability	Cluster Match probability for an Incident with its top cluster as determined by the Incident Clustering model. This attribute is formatted in the iCube to display values in percentage format with one decimal point.
Associated Problem Number	Problem Number associated with an Incident or Cluster
Associated Change Number	Change Number associated with an Incident or Cluster
Incident	Unique identifier to identify an incident
Incident Closed Date	Date on which the incident was closed
Incident Number	Incident number as captured in the source system
Incident Opened Date	Date on which the incident was opened
Incident Priority	Priority of the incident as mentioned in the source system
Incident Problem	Problem associated with an Incident or Cluster
Incident Short Description	Incident short description as captured in the source system
Model Build Date	Date on which the clustering model was built
Month	Gregorian calendar month displayed as "Mmm YYYY"

Metrics

Expand this link to view the Incident Cluster metrics, their descriptions, and their formula.

Metric Name	Description	Formula	Expected Value
#1 Cluster Match Probability	Displays the Cluster Match probability for an Incident with its top cluster as determined by the Incident Clustering model. This is a derived metric based on the #1 Cluster Match Probability attribute and is used to create a dashboard level slider filter.	Max [#1 Cluster Match Probability (attribute)]	>=0
% Change MoM	Month-on-month percentage change in the volume of Incidents for a particular Cluster	([Opened Incidents] - [Opened Incidents of Previous Month]) / [Opened Incidents of Previous Month])	>0% or <0% Ideal value would be <0%
% Change MoM Rank	Rank of a cluster based on the month-on-month percentage change in the volume of Incidents; relevant for a Cluster and Month combination	Rank ([% Change MoM])	>0
Cluster % to Opened	Displays incidents belonging to a cluster as a % of Opened Incidents. Note: This metric is calculated only at the #1 Cluster level and is only available from the Incident Cluster iCube.	[Opened Incidents belonging to #1 cluster] / Sum ([Opened Incidents])	0% to 100%
Incidents with >40% Cluster Match	Count of Incidents whose top Cluster Match probability is greater than 40%	Count (Opened Incidents) WHERE Cluster 1 Match Probability is Greater than 40	>=0
Linked KB articles	Count of knowledge base articles associated with an Incident or Cluster	Sum (Linked KB Articles)	>=0
Opened Incidents	Count of Incidents, which are opened during the cube time filter	Count of all Incidents	>=0
Resolved Incidents	Count of Incidents which have been resolved during the cube time filter	Count of all Incidents WHERE Incident State is (RESOLVED, CLOSED)	>=0
Total Clusters	Count of total Clusters that were identified after clustering	Count of all Clusters	>=0
Total Incidents	Count of total Incidents that were used for clustering. This is a derived metric and is created only within the Incident Cluster iCube.	Sum (Opened Incidents)	>=0
Volume Change Mom	Month-on-month change in the volume of Incidents for a particular Cluster	([Opened Incidents] - [Opened Incidents of Previous Month])	<0, >0, or =0
Volume Change MoM Rank	Rank of a cluster based on the month-on-month changes in the volume of Incidents; relevant for a Cluster and Month combination	Rank ([Volume Change MoM])	>0

Incident Clusters Description dataset

Enables you to create reports that provide information about clusters, keywords associated with the clusters, weightage for each keyword, and cluster descriptions. This dataset can be blended with the Incident Cluster dataset to derive cluster-level insights for Incidents.

Attributes

Expand this link to view the Incident Clusters Description attributes:

Attributes	Description
Cluster	Unique number or identifier of a Cluster as defined by the Clustering algorithm
Cluster Description	Comma separated concatenation of keywords associated with a Cluster Note: Only the top five keywords are displayed in the descending order of keyword weight.
Keyword	Individual keyword that is associated with a Cluster
Keyword Weight	Weight of the keyword in a particular Cluster as identified by the Clustering algorithm