Trait CvSplitter

Source

pub trait CvSplitter {
    // Required methods
    fn split(&self, n_samples: usize) -> Result<Vec<(Vec<usize>, Vec<usize>)>>;
    fn n_splits(&self) -> usize;
}

Expand description

A cross-validation splitting strategy.

Implementors return (train, test) index pairs for a dataset of n_samples rows. The indices are into the original row order; apply them with ndarray::ArrayBase::select.

use model_selection_rs::splitters::{CvSplitter, KFold};

let kf = KFold::new(3).unwrap();
let splits = kf.split(6).unwrap();
assert_eq!(splits.len(), 3);
for (train, test) in &splits {
    assert_eq!(train.len() + test.len(), 6);
}

Required Methods§

Source

fn split(&self, n_samples: usize) -> Result<Vec<(Vec<usize>, Vec<usize>)>>

Produce every (train_indices, test_indices) pair for n_samples rows.

§Errors

Returns ModelSelectionError if the configuration cannot produce valid splits for n_samples (for example more folds than samples, or — for stored-label splitters — an n_samples that disagrees with the stored label array length).

Source