Struct pelt::Pelt

pub struct Pelt { /* fields omitted */ }

Pruned Exact Linear Time changepoint detection model.

Arguments

• jump - Distance between proposal changepoints.
• min_size - Minimal distance between changepoints.
• n_samples - No. of samples in signal.
• loss - Loss function to use.
• pen - Penalty term.

Example

let mut m = Pelt::new(jump, min_size, loss, pen);
println!("{:?}", m.predict(&signal))

Methods

impl Pelt

pub fn new(
    jump: Option<usize>,
    min_size: Option<usize>,
    loss: Option<&str>,
    pen: f64
) -> Pelt

Create new Pelt instance. Note that the variable should be created mutable.

Trait Implementations

impl MutEstimator<Vec<usize>> for Pelt

fn fit(&mut self, signal: &Vec2d) -> &Self

Set the number of samples in the signal. Explicit call is not needed.

Panics

if min_size < signal.len()

fn predict(&mut self, signal: &Vec2d) -> Option<Vec<usize>>

Retreive the changepoints in the signal.

Arguments

• signal - Signals stacked in 2D. All signals should be equal length.

Panics

if min_size < signal.len()

fn fit_predict(&mut self, signal: &Vec2d) -> Option<Vec<usize>>

See predict