Auto Cluster Sets — JWGeotech

Configure deterministic clustering, review suggested joint sets, and apply assignments.

Auto Cluster Sets uses fuzzy clustering on the orientations in the current filter. Accepted solver poles fit ordinary editable assignment windows. The preview applies those windows to the entire dataset, including filtered-out poles and solver exclusions that fall inside the fitted geometry. Previewed membership can therefore differ from fuzzy classifications.

Cluster Inputs

Input	Range or values	Meaning
Minimum Sets	2 to 20	Lowest number of candidate clusters to evaluate.
Maximum Sets	2 to 20	Highest number of candidate clusters to evaluate.
Minimum Membership (%)	0 to 100	Membership threshold for confident assignment.
Confidence Interval (%)	1 to 100	Angular confidence interval used in review.
Orientation Vector	Pole Vector, Dip Vector	Vector representation used for clustering.
Displayed Sets	Generated after preview	Selects which candidate cluster count is displayed.
Cluster unassigned poles only	On/off	Uses only unassigned poles in the current filter to fit clusters. Generated windows still apply globally.
Acceptance	Add new sets (default), Replace all sets	Preserve existing definitions with unique new names, or replace all assignment definitions.
Preview Cluster Sets	Button	Runs the clustering preview.

Cluster Review Actions

Action	What it does
Accept All	Commits the previewed definitions and windows for all selected proposals.
Accept Selected Sets	Commits the definitions and windows shown for the selected proposals.
Undo	Discards an active preview, or restores the definitions, windows and assignments from before acceptance.

Cluster Outputs

Output	Meaning
Variant summary	Shows evaluated set-count range and selected candidate.
Proposed Sets	Shows each set's selection toggle, label, actual resulting orientation count and mean plane, plus the original solver confidence interval angle. It stays a table at the 430 px sidebar minimum; at narrower pane widths, each set becomes a labeled stacked review row.
Cluster preview on plot	Draws preview assignments before table labels are committed.

Window fitting is deterministic. It handles north and equator wrapping, splits groups wider than the supported 180° window span, and gives coincident or single poles a minimum 1° extent. Accepted windows can be resized or deleted in Set Assignment.