Regions

Combines the branches extracted by Trees and prioritizes those that form consistent clusters.

Main functions

• prio_ranges(branches, X, y=None): computes coverage, purity, and priority metrics.
• labels(X, ranges, return_descriptions=False): assigns labels and optionally generates descriptions.
• plot_multi_dims(range_df, data, target): visualizes regions across multiple dimensions.

Usage

from InsideForest import Regions

regions = Regions()
priority_ranges = regions.prio_ranges(branches, X)
clusterized, descriptions = regions.labels(X, priority_ranges, True)

Relevant parameters

• min_support: minimum region size to be considered.
• max_rules: maximum number of rules per cluster.
• use_probabilities: weights by forest probabilities.

The output of labels feeds into Labels and the methods in InsideForestRegionClusterer.