Struct ClusterSupportingObjects

pub struct ClusterSupportingObjects { /* private fields */ }

Implementations

impl ClusterSupportingObjects

pub fn count(&self) -> usize

Returns the number of cluster supporting objects.

Examples found in repository

examples/sample.rs (line 67)

fn main() -> io::Result<()> {
    let filename = env::args().nth(1);
    if filename.is_none() {
        eprintln!("No input file");
        exit(1);
    }
    let data = read_data(io::BufReader::new(File::open(filename.unwrap())?))?;
    let flame = DistanceGraph::build(&data, distance::euclidean);

    print!("Detecting Cluster Supporting Objects ...");
    io::stdout().flush()?;
    let supports = flame.find_supporting_objects(10, -2.0);
    println!("done, found {}", supports.count());

    print!("Propagating fuzzy memberships ... ");
    io::stdout().flush()?;
    let fuzzyships = supports
        .approximate_fuzzy_memberships(500, 1e-6)
        .assign_outliers();
    println!("done");

    print!("Defining clusters from fuzzy memberships ... ");
    io::stdout().flush()?;
    let (clusters, outliers) = fuzzyships.make_clusters(-1.0);
    println!("done");

    for (i, cluster) in clusters.iter().enumerate() {
        print!("\nCluster {:3}, with {:6} members:\n", i + 1, cluster.len());
        print_cluster(cluster);
    }
    print!("\nCluster outliers, with {:6} members:\n", outliers.len());
    print_cluster(&outliers);

    Ok(())
}

pub fn object_type(&self, id: usize) -> ObjectType

Identifies how an object was classified by the algorithm.

pub fn approximate_fuzzy_memberships(self, steps: usize, epsilon: f64) -> Self

Local Approximation of fuzzy memberships. Stopped after the maximum steps of iterations; Or stopped when the overall membership difference between two iterations become less than epsilon.

Examples found in repository

examples/sample.rs (line 72)

fn main() -> io::Result<()> {
    let filename = env::args().nth(1);
    if filename.is_none() {
        eprintln!("No input file");
        exit(1);
    }
    let data = read_data(io::BufReader::new(File::open(filename.unwrap())?))?;
    let flame = DistanceGraph::build(&data, distance::euclidean);

    print!("Detecting Cluster Supporting Objects ...");
    io::stdout().flush()?;
    let supports = flame.find_supporting_objects(10, -2.0);
    println!("done, found {}", supports.count());

    print!("Propagating fuzzy memberships ... ");
    io::stdout().flush()?;
    let fuzzyships = supports
        .approximate_fuzzy_memberships(500, 1e-6)
        .assign_outliers();
    println!("done");

    print!("Defining clusters from fuzzy memberships ... ");
    io::stdout().flush()?;
    let (clusters, outliers) = fuzzyships.make_clusters(-1.0);
    println!("done");

    for (i, cluster) in clusters.iter().enumerate() {
        print!("\nCluster {:3}, with {:6} members:\n", i + 1, cluster.len());
        print_cluster(cluster);
    }
    print!("\nCluster outliers, with {:6} members:\n", outliers.len());
    print_cluster(&outliers);

    Ok(())
}

pub fn assign_outliers(self) -> Self

Attempt to assign outliers to the nearest cluster.

Examples found in repository

examples/sample.rs (line 73)

fn main() -> io::Result<()> {
    let filename = env::args().nth(1);
    if filename.is_none() {
        eprintln!("No input file");
        exit(1);
    }
    let data = read_data(io::BufReader::new(File::open(filename.unwrap())?))?;
    let flame = DistanceGraph::build(&data, distance::euclidean);

    print!("Detecting Cluster Supporting Objects ...");
    io::stdout().flush()?;
    let supports = flame.find_supporting_objects(10, -2.0);
    println!("done, found {}", supports.count());

    print!("Propagating fuzzy memberships ... ");
    io::stdout().flush()?;
    let fuzzyships = supports
        .approximate_fuzzy_memberships(500, 1e-6)
        .assign_outliers();
    println!("done");

    print!("Defining clusters from fuzzy memberships ... ");
    io::stdout().flush()?;
    let (clusters, outliers) = fuzzyships.make_clusters(-1.0);
    println!("done");

    for (i, cluster) in clusters.iter().enumerate() {
        print!("\nCluster {:3}, with {:6} members:\n", i + 1, cluster.len());
        print_cluster(cluster);
    }
    print!("\nCluster outliers, with {:6} members:\n", outliers.len());
    print_cluster(&outliers);

    Ok(())
}

pub fn make_clusters(&self, thd: f64) -> (Vec<Vec<usize>>, Vec<usize>)

Construct clusters. If 0<thd<1: each object is assigned to all clusters in which it has membership higher than thd; if it can not be assigned to any clusters, it is then assigned to the outlier group. Else: each object is assigned to the group (clusters/outlier group) in which it has the highest membership.

Examples found in repository

examples/sample.rs (line 78)

fn main() -> io::Result<()> {
    let filename = env::args().nth(1);
    if filename.is_none() {
        eprintln!("No input file");
        exit(1);
    }
    let data = read_data(io::BufReader::new(File::open(filename.unwrap())?))?;
    let flame = DistanceGraph::build(&data, distance::euclidean);

    print!("Detecting Cluster Supporting Objects ...");
    io::stdout().flush()?;
    let supports = flame.find_supporting_objects(10, -2.0);
    println!("done, found {}", supports.count());

    print!("Propagating fuzzy memberships ... ");
    io::stdout().flush()?;
    let fuzzyships = supports
        .approximate_fuzzy_memberships(500, 1e-6)
        .assign_outliers();
    println!("done");

    print!("Defining clusters from fuzzy memberships ... ");
    io::stdout().flush()?;
    let (clusters, outliers) = fuzzyships.make_clusters(-1.0);
    println!("done");

    for (i, cluster) in clusters.iter().enumerate() {
        print!("\nCluster {:3}, with {:6} members:\n", i + 1, cluster.len());
        print_cluster(cluster);
    }
    print!("\nCluster outliers, with {:6} members:\n", outliers.len());
    print_cluster(&outliers);

    Ok(())
}

Trait Implementations

impl Debug for ClusterSupportingObjects

fn fmt(&self, __f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.