Trait abd_clam::Cluster

pub trait Cluster<U: Number>: Serialize + for<'a> Deserialize<'a> + PartialEq + Eq + PartialOrd + Ord + Debug + Hash + Display + Send + Sync + Clone {
    // Required methods
    fn new_root<I: Instance, D: Dataset<I, U>>(
        data: &D,
        seed: Option<u64>,
    ) -> Self;
    fn partition<I, D, P>(
        self,
        data: &mut D,
        criteria: &P,
        seed: Option<u64>,
    ) -> Self
       where I: Instance,
             D: Dataset<I, U>,
             P: PartitionCriterion<U>;
    fn offset(&self) -> usize;
    fn cardinality(&self) -> usize;
    fn depth(&self) -> usize;
    fn arg_center(&self) -> usize;
    fn radius(&self) -> U;
    fn arg_radial(&self) -> usize;
    fn lfd(&self) -> f64;
    fn children(&self) -> Option<[&Self; 2]>;
    fn polar_distance(&self) -> Option<U>;
    fn arg_poles(&self) -> Option<[usize; 2]>;

    // Provided methods
    fn name(&self) -> String { ... }
    fn descend_to(&self, offset: usize, cardinality: usize) -> Option<&Self> { ... }
    fn is_ancestor_of(&self, other: &Self) -> bool { ... }
    fn is_descendant_of(&self, other: &Self) -> bool { ... }
    fn is_leaf(&self) -> bool { ... }
    fn is_singleton(&self) -> bool { ... }
    fn indices(&self) -> Range<usize> { ... }
    fn subtree(&self) -> Vec<&Self> { ... }
    fn max_leaf_depth(&self) -> usize { ... }
    fn distance_to_instance<I: Instance, D: Dataset<I, U>>(
        &self,
        data: &D,
        instance: &I,
    ) -> U { ... }
    fn distance_to_other<I: Instance, D: Dataset<I, U>>(
        &self,
        data: &D,
        other: &Self,
    ) -> U { ... }
    fn overlapping_children<I: Instance, D: Dataset<I, U>>(
        &self,
        data: &D,
        query: &I,
        radius: U,
    ) -> Vec<&Self> { ... }
    fn save(&self, path: &Path) -> Result<(), String> { ... }
    fn load(path: &Path) -> Result<Self, String> { ... }
}

A Cluster represents a set of “similar” instances under some distance function.

Required Methods

fn new_root<I: Instance, D: Dataset<I, U>>(data: &D, seed: Option<u64>) -> Self

Creates a new Cluster from a given dataset.

fn partition<I, D, P>( self, data: &mut D, criteria: &P, seed: Option<u64>, ) -> Self

where I: Instance, D: Dataset<I, U>, P: PartitionCriterion<U>,

Recursively partitions the Cluster until the PartitionCriteria are met.

fn offset(&self) -> usize

The offset of the indices of the Cluster’s instances in the dataset.

fn cardinality(&self) -> usize

The number of points in the cluster.

fn depth(&self) -> usize

The depth of the cluster in the tree.

fn arg_center(&self) -> usize

The index of the instance at the center of the Cluster.

TODO: Remove this method when we have “center-less” clusters.

fn radius(&self) -> U

The radius of the cluster.

fn arg_radial(&self) -> usize

The index of the instance with the maximum distance from the center

fn lfd(&self) -> f64

The local fractal dimension of the å.

fn children(&self) -> Option<[&Self; 2]>

The two child clusters.

fn polar_distance(&self) -> Option<U>

The distance between the two poles of the Cluster used for partitioning.

fn arg_poles(&self) -> Option<[usize; 2]>

The indices of the instances used as poles for partitioning.

Provided Methods

fn name(&self) -> String

The name of the Cluster String.

This is a human-readable representation of the Cluster’s offset and cardinality. It is a unique identifier in the tree. It may be used to store the Cluster in a database, or to identify the Cluster in a visualization.

fn descend_to(&self, offset: usize, cardinality: usize) -> Option<&Self>

Descends to the Cluster with the given offset and cardinality.

If such a Cluster does not exist, None is returned.

Arguments

• offset: The offset of the Cluster’s instances in the dataset.
• cardinality: The number of instances in the Cluster.

fn is_ancestor_of(&self, other: &Self) -> bool

Whether the Cluster is an ancestor of another Cluster.

fn is_descendant_of(&self, other: &Self) -> bool

Whether the Cluster is a descendant of another Cluster.

fn is_leaf(&self) -> bool

Whether the Cluster is a leaf node in the tree.

fn is_singleton(&self) -> bool

Whether the Cluster is a singleton, i.e. it contains only one instance or has a radius of zero.

fn indices(&self) -> Range<usize>

The indices of the instances in the Cluster after the dataset has been reordered.

fn subtree(&self) -> Vec<&Self>

The subtree of the Cluster.

fn max_leaf_depth(&self) -> usize

The maximum depth of and leaf in the subtree of the Cluster.

If this Cluster is a leaf, the maximum depth is the depth of the Cluster.

fn distance_to_instance<I: Instance, D: Dataset<I, U>>( &self, data: &D, instance: &I, ) -> U

Distance from the center to the given instance.

fn distance_to_other<I: Instance, D: Dataset<I, U>>( &self, data: &D, other: &Self, ) -> U

Distance from the center of this Cluster to the center of the other Cluster.

fn overlapping_children<I: Instance, D: Dataset<I, U>>( &self, data: &D, query: &I, radius: U, ) -> Vec<&Self>

Assuming the Cluster overlaps with the query ball, we return only those children that also overlap with the query ball.

fn save(&self, path: &Path) -> Result<(), String>

Saves a Cluster to a given location.

Arguments

• path: The path to the Cluster file.

Errors

• If the file cannot be created.
• If the file cannot be serialized.

fn load(path: &Path) -> Result<Self, String>

Loads a Cluster from a given location.

Arguments

• path: The path to the Cluster file.

Returns

• The Cluster loaded from the file.

Errors

• If the file cannot be opened.
• If the file cannot be deserialized.

Object Safety

This trait is not object safe.

Implementors

impl<U: Number> Cluster<U> for Vertex<U>

impl<U: Number> Cluster<U> for UniBall<U>

impl<U: UInt> Cluster<U> for SquishyBall<U>