Struct Ewma 

pub struct Ewma { /* private fields */ }

Exponentially weighted moving average (EWMA).

α ∈ (0, 1] is the smoothing factor. Higher α gives more weight to recent observations. The first call to update sets the initial value to the first sample (no warm-up bias).

Usage

use oximedia_transcode::running_stats::Ewma;

let mut ewma = Ewma::new(0.2);
for v in [100.0_f64, 110.0, 90.0, 105.0] {
    ewma.update(v);
}
// value is close to the true mean but smoothed
assert!((ewma.value() - 100.0).abs() < 20.0);

Implementations

impl Ewma

pub fn new(alpha: f64) -> Self

Creates an EWMA with smoothing factor alpha.

Panics

Panics in debug builds when alpha is outside (0.0, 1.0].

pub fn update(&mut self, sample: f64)

Updates the EWMA with a new observation.

Silently ignores NaN values.

pub fn value_opt(&self) -> Option<f64>

Returns the current EWMA value, or None before any update.

pub fn value(&self) -> f64

Returns the current EWMA value, or 0.0 if not yet initialised.

pub fn variance(&self) -> f64

Returns the EWMA-based variance estimate.

pub fn stddev(&self) -> f64

Returns the EWMA-based standard-deviation estimate.

pub fn count(&self) -> u64

Returns the number of observations processed.

pub fn alpha(&self) -> f64

Returns the smoothing factor α.

pub fn reset(&mut self)

Resets the EWMA to its initial state (no observations).

Trait Implementations

impl Clone for Ewma

fn clone(&self) -> Ewma

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Ewma

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

Formats the value using the given formatter.