SHCEvalCallback: Statistical Hierarchical Clusterer Evaluation Callbacks

In dkrleza/SHClus: Statistical Hierarchical Clustering

Description

Allows SHC evaluation callback for the stream package. This call contains measures mostly related to use of the sigma-index.

Fields

all_measures

The following list of measures are available:

• queryTime - Total query time for the last processing.

• updateTime - Total sigma-index update time.

• processTime - Total time needed for the last processing... measured inside SHC algorithm.

• nodeCount - Total mahalanobis distances calculated for the last processing.

• computationCostReduction - Computational cost reduction when using sigma-index comparing to the sequential scan.

internal_measures

Same as all_measures.

external_measures

Empty.

outlier_measures

Empty.

Author(s)

Dalibor Krleža

Examples

set.seed(0)
d <- DSD_Gaussians(k = 3, d = 2, outliers = 5, separation_type = "Mahalanobis", 
      separation = 4, space_limit = c(0, 25), variance_limit = 1,
      outlier_options = list(outlier_horizon = 10000))
c <- DSC_SHC.man(2, 3.2, 0.3, cbNLimit = 0, cbVarianceLimit = 0,decaySpeed = 0,
      sigmaIndex = TRUE)
evaluate_with_callbacks(c, d, n = 10000, type = "macro", 
      measure = c("crand", "outlierjaccard", "computationCostReduction"), 
      callbacks = list(shc = SHCEvalCallback()))

dkrleza/SHClus documentation built on Feb. 25, 2021, 10:30 p.m.