CGAData{S, T<:Real}

Contains the data used in the computation of cga.

The user does not need to query this object, but can use this as an opaque object. The results of the computation can be obtained from CGAResult object.

CGAResult <: ClusteringResult

Contains the results from the computation of cga.

Members

• assignments::Vector{Int} element-to-cluster assignments (n)
• counts::Vector{Int} number of samples assigned to each cluster (k)
• found_gen::Int first generation where the elite was found
• total_gen::Int total generations the GA has been run

cga(objects::V,
    distances::M=distance_matrix(objs),
    N::Int=length(objs)*20,
    generations::Int=50) where {S, V <: AbstractVector{S},
                                T <: Real, M <: AbstractMatrix{T}}

Compute the optimal clustering in the data by Genetic Algorithm over the computed mean silhouettes.

• objects the vector of the objects for which clusters are to be computed.
• distances the distance matrix providing the pairwise distances bewtween the objects
• N the population size for GA computation
• generations number of generations the GA has to be run

The fitness function used is 1+mean(silhouettes()) to ensure positive values.

Return Values

The method returns a tuple of (CGAData, CGAResult)

References

1. Hruschka, Eduardo & Ebecken, Nelson. (2003). A genetic algorithm for cluster analysis. Intell. Data Anal.. 7. 15-25. 10.3233/IDA-2003-7103.