Struct MinMaxScaler 

pub struct MinMaxScaler { /* private fields */ }

Streaming min-max scaler – incremental feature scaling to a configurable range.

Tracks per-feature running minimum and maximum values, scaling features to a target range (default [0, 1]). Each call to update_and_transform incorporates a new sample in O(n_features) time with zero allocations (after the first sample initializes internal buffers).

Scaling formula:

scaled = (x - min) / (max - min) * (range_high - range_low) + range_low

When max - min < f64::EPSILON for a feature (zero range), the scaler outputs the midpoint of the target range to avoid division by zero.

Example

use irithyll::preprocessing::MinMaxScaler;
use irithyll::pipeline::StreamingPreprocessor;

let mut scaler = MinMaxScaler::new();

// Feed samples one at a time.
scaler.update_and_transform(&[0.0, 100.0]);
scaler.update_and_transform(&[10.0, 0.0]);

// Transform a new observation using running min/max.
let scaled = scaler.transform(&[5.0, 50.0]);
assert!((scaled[0] - 0.5).abs() < 1e-9);
assert!((scaled[1] - 0.5).abs() < 1e-9);

Implementations

impl MinMaxScaler

pub fn new() -> Self

Create a scaler with the default target range [0, 1].

The number of features is inferred from the first call to update_and_transform.

use irithyll::preprocessing::MinMaxScaler;
let scaler = MinMaxScaler::new();

pub fn with_range(low: f64, high: f64) -> Self

Create a scaler with a custom output range [low, high].

Panics

Panics if low >= high.

use irithyll::preprocessing::MinMaxScaler;
let scaler = MinMaxScaler::with_range(-1.0, 1.0);

pub fn mins(&self) -> &[f64]

Per-feature running minimums observed so far.

Empty until the first sample has been incorporated.

pub fn maxs(&self) -> &[f64]

Per-feature running maximums observed so far.

Empty until the first sample has been incorporated.

pub fn range(&self) -> (f64, f64)

The configured output range as (low, high).

pub fn count(&self) -> u64

Number of samples incorporated so far.

Trait Implementations

impl Clone for MinMaxScaler

fn clone(&self) -> MinMaxScaler

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MinMaxScaler

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

Formats the value using the given formatter.

impl Default for MinMaxScaler

fn default() -> Self

Returns the “default value” for a type.

impl StreamingPreprocessor for MinMaxScaler

fn update_and_transform(&mut self, features: &[f64]) -> Vec<f64>

Update running min/max from this sample and return scaled features.

On the first call, initializes internal buffers from the sample. Since min == max for every feature after a single sample, all outputs will be the midpoint of the target range.

fn transform(&self, features: &[f64]) -> Vec<f64>

Scale features using current min/max without updating statistics.

If no samples have been seen yet (no min/max stats), features pass through unscaled.

fn output_dim(&self) -> Option<usize>

Number of output features, or None if unknown until the first sample.

fn reset(&mut self)

Reset to initial (untrained) state.

Clears mins, maxs, sets n_features to None, and resets count to 0. Preserves the configured output range.