Struct Msnn 

pub struct Msnn { /* private fields */ }

A blend of local linear models with partition-of-unity triangular memberships over one scheduling input.

Implementations

impl Msnn

pub fn membership(&self, x: f64) -> Vec<f64>

The normalized membership weights at x (they sum to exactly 1 — the partition of unity).

pub fn fit(xs: &[f64], ys: &[f64], centers: usize, lo: f64, hi: f64) -> Msnn

Fit centers local linear models on [lo, hi]. Each local model i is an independent membership-weighted linear regression, so its slope aᵢ is a well-defined, readable local slope; the prediction blends them by the partition-of-unity memberships.

pub fn predict(&self, x: f64) -> f64

Predict f(x) = blend of the local linear models.

pub fn n_centers(&self) -> usize

pub fn center(&self, i: usize) -> f64

pub fn local_slope(&self, i: usize) -> f64

The learned local slope of model i — the function’s derivative near center(i).

pub fn local_line(&self, i: usize, x: f64) -> f64

The local line aᵢ x + bᵢ evaluated at x (for drawing the individual pieces).