BIRCHOptions<T>

Options & Configuration · AiDotNet.Clustering.Options

Configuration options for BIRCH (Balanced Iterative Reducing and Clustering using Hierarchies).

Inheritance: ModelOptions → ClusteringOptions<T> → BIRCHOptions<T>

For Beginners

BIRCH is like creating a summary tree of your data.

Imagine organizing a library:

  1. First, create sections (CF-tree nodes)
  2. Group similar books together (clustering features)
  3. Each node summarizes: count, sum, sum of squares
  4. Finally, cluster the summaries

Benefits:

  • Handles very large datasets efficiently
  • Incremental: can add data without rebuilding
  • Single pass through data (mostly)
  • Memory efficient with controllable tree size

Key parameters:

  • Threshold: How similar points must be to join a cluster
  • BranchingFactor: Max children per node (controls tree width)

How It Works

BIRCH incrementally builds a CF (Clustering Feature) tree to summarize the data, then applies a clustering algorithm to the leaf entries. It’s designed for very large datasets that don’t fit in memory.

Constructors

Constructor Summary
BIRCHOptions() Initializes BIRCHOptions with appropriate defaults.

Properties

Property Summary
bool ComputeLabels { get; set; } Gets or sets whether to compute cluster labels.
bool Copy { get; set; } Gets or sets whether to copy the input data.
double Threshold { get; set; } Gets or sets the threshold for cluster radius.
int BranchingFactor { get; set; } Gets or sets the maximum branching factor.
int? NumClusters { get; set; } Gets or sets the target number of clusters.