Clusters: private API
PortfolioOptimisers.AbstractClustersEstimator — Type
abstract type AbstractClustersEstimator <: AbstractPhylogenyEstimatorAbstract supertype for all clustering estimator types.
A subtype holds the configuration that clusters the assets of a universe. clusterise runs it and returns an AbstractClusteringResult. ClustersEstimator and NetworkClustersEstimator are the two subtypes.
Related
PortfolioOptimisers.AbstractClustersAlgorithm — Type
abstract type AbstractClustersAlgorithm <: AbstractPhylogenyAlgorithmAbstract supertype for all clustering algorithm types.
A subtype selects the method that clusters a distance matrix. The hierarchical algorithms and the non-hierarchical algorithms are the two families.
Related
PortfolioOptimisers.AbstractHierarchicalClusteringAlgorithm — Type
abstract type AbstractHierarchicalClusteringAlgorithm <: AbstractClustersAlgorithmAbstract supertype for all hierarchical clustering algorithm types.
A subtype builds a dendrogram, a Clustering.Hclust. optimal_number_clusters then cuts the dendrogram at the selected number of clusters. HClustAlgorithm and DBHT are the subtypes.
Related
PortfolioOptimisers.AbstractNonHierarchicalClusteringAlgorithm — Type
abstract type AbstractNonHierarchicalClusteringAlgorithm <: AbstractClustersAlgorithmAbstract supertype for all non-hierarchical clustering algorithm types.
A subtype makes a flat partition, a Clustering.ClusteringResult, at one number of clusters. A partition has no dendrogram to cut, so optimal_number_clusters runs the algorithm once for each candidate count. KMeansAlgorithm is the subtype.
Related
PortfolioOptimisers.AbstractOptimalNumberClustersEstimator — Type
abstract type AbstractOptimalNumberClustersEstimator <: AbstractEstimatorAbstract supertype for all optimal number of clusters estimator types.
A subtype holds the configuration that selects the number of clusters. OptimalNumberClusters is the subtype.
Related
PortfolioOptimisers.AbstractOptimalNumberClustersAlgorithm — Type
abstract type AbstractOptimalNumberClustersAlgorithm <: AbstractAlgorithmAbstract supertype for all optimal number of clusters algorithm types.
A subtype is a score that optimal_number_clusters calculates for each candidate count, and the count with the largest score wins. SecondOrderDifference and SilhouetteScore are the subtypes.
Related
PortfolioOptimisers.AbstractClusteringResult — Type
abstract type AbstractClusteringResult <: AbstractPhylogenyResultAbstract supertype for all clustering result types.
A subtype holds the output of a clustering estimator. Clusters is the subtype.
Related
PortfolioOptimisers.Int_ONC — Type
const Int_ONC = Union{Integer, AbstractOptimalNumberClustersAlgorithm}Alias for an integer or optimal number of clusters algorithm.
Matches an integer, which states the number of clusters, or an AbstractOptimalNumberClustersAlgorithm, which selects the number from the data.
Related
PortfolioOptimisers.ClTypes — Type
const ClTypes = Union{Clustering.ClusteringResult, Clustering.Hclust}Alias for clustering result types from the Clustering.jl package.
Matches a Clustering.ClusteringResult or a Clustering.Hclust. The res field of Clusters accepts either, so one result type holds a flat partition or a dendrogram.
Related
PortfolioOptimisers.ClE_Cl — Type
const ClE_Cl = Union{AbstractClustersEstimator, AbstractClusteringResult}Alias for a clustering estimator or result.
Matches an AbstractClustersEstimator or an AbstractClusteringResult. The library dispatches on it where a function accepts a clustering estimator or a precomputed result.
Related