The Issue Classification dashboard provides an in-depth way to find similar issues and possible duplicates by grouping similar items in a cluster relationship. Once grouped, issues in the clusters can be further investigated.
The bubble chart below displays clusters in a given time range. Items in a cluster range from somewhat similar to mutually similar; the degree of similarity is indicated by the color of the bubble.
The selected cluster above includes two issues, #smb-148 and #smb-152. Click an issue to reveal how it is related to other issues. As shown above, the dashboard returns the following information for issue #smb-148:
- the similar issue (smb-152)
- matching attributes (matches) or terms that are common to both smb-148 and smb-152
- a rank,which indicates how relevant smb-152 is to smb-148
- a score, which indicates how relevant smb-152 is to the cluster
Group Details shows supporting data that issues in a cluster share which makes them similar. As you click a different cluster, items in Group Details change.
Clusters in the example are based on a Similarity Score Cutoff of .33 which indicates a more general relationship among issues. Adjust the Similarity Score Cutoff to reveal isolated clusters. A lower cutoff score (0.1) will show relationships among all items on a general level which might not be meaningful.
All information can be saved to a CSV file by clicking Export CSV.