Struct QuantileNormF32 

pub struct QuantileNormF32 { /* private fields */ }

Quantile normalizer via P² estimator grid.

Maps values to approximate uniform [0, 1] by maintaining a grid of P² percentile estimators at uniformly spaced quantile points. The normalized value is the interpolated rank within the grid.

O(resolution) per update. Requires alloc.

Examples

use nexus_stats_core::normalization::QuantileNormF32;

let mut qn = QuantileNormF32::builder().resolution(9).build().unwrap();
for i in 0..500 {
    let _ = qn.update(i as f32);
}
let v = qn.update(250.0).unwrap();
assert!(v.is_some());

Implementations

impl QuantileNormF32

pub fn builder() -> QuantileNormF32Builder

Creates a builder.

pub fn update(&mut self, sample: f32) -> Result<Option<f32>, DataError>

Feeds a sample to all grid estimators. Returns the approximate normalized rank in [0, 1] once primed.

Errors

Returns DataError::NotANumber if the sample is NaN, or DataError::Infinite if the sample is infinite.

pub fn normalize(&self, value: f32) -> Option<f32>

Ranks an arbitrary value against current grid without updating state.

Returns None if not primed.

pub fn resolution(&self) -> usize

Number of quantile grid points.

pub fn count(&self) -> u64

Total samples processed.

pub fn is_primed(&self) -> bool

Whether all grid estimators are primed.

pub fn reset(&mut self)

Resets all grid estimators. Resolution preserved.

Trait Implementations

impl Clone for QuantileNormF32

fn clone(&self) -> QuantileNormF32

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for QuantileNormF32

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

Formats the value using the given formatter.