Trait indxvec::Indices

pub trait Indices {
    // Required methods
    fn invindex(self) -> Vec<usize>;
    fn complindex(self) -> Vec<usize>;
    fn unindex<T>(self, v: &[T], ascending: bool) -> Vec<T>
       where T: Clone;
    fn ucorrelation(self, v: &[usize]) -> f64;
    fn indx_to_f64(self) -> Vec<f64>;

    // Provided method
    fn newindex(n: usize) -> Vec<usize> { ... }
}

Methods to manipulate indices of Vec<usize> type.

Required Methods

fn invindex(self) -> Vec<usize>

Invert an index - turns a sort order into rank order and vice-versa

fn complindex(self) -> Vec<usize>

complement of an index - reverses the ranking order

fn unindex<T>(self, v: &[T], ascending: bool) -> Vec<T>

where T: Clone,

Collect values from v in the order of indices in self.

fn ucorrelation(self, v: &[usize]) -> f64

Correlation coefficient of two &usize slices. Pearsons on raw data, Spearman’s when applied to ranks.

fn indx_to_f64(self) -> Vec<f64>

Potentially useful clone-recast of &usize to Vec

Provided Methods

fn newindex(n: usize) -> Vec<usize>

Indices::newindex(n) creates a new index without rePartialOrdering

Object Safety

This trait is not object safe.

Implementations on Foreign Types

impl Indices for &[usize]

fn invindex(self) -> Vec<usize>

Inverts an index, eg. from sort index to ranks. This is a symmetric operation: any even number of applications gives the original index, odd number gives the inverted form.

fn unindex<T>(self, v: &[T], ascending: bool) -> Vec<T>

where T: Clone,

Collects values from v in the order given by self index. When ascending is false, collects in descending order.
It is used here by msort for ascending or descending sort.

fn complindex(self) -> Vec<usize>

Complement of an index (is symmetric) - .complindex() toggles rank index between ascending/descending. To toggle sort index between ascending/descending, use the general reversal revs: ranks.complindex().invindex() = ranks.invindex().revs()

fn ucorrelation(self, v: &[usize]) -> f64

Pearson’s correlation coefficient of two $[usize] slices. When the inputs are ranks, then this gives Spearman’s correlation of the original data. However, in general, any other ordinal measures could be deployed (not just the ranks).

fn indx_to_f64(self) -> Vec<f64>

Potentially useful clone-recast of &usize to Vec

Implementors