Function label_encode 

pub fn label_encode<T: Clone + Ord>(
    labels: &Array1<T>,
) -> Result<(Array1<usize>, Vec<T>), DatasetError>

Map labels of any type to consecutive integer codes.

Turns a label vector into the 0..n_classes codes most training code expects. A loader holds its labels in an Array1<String>, and this function also accepts any other comparable type. It returns the class list needed to decode a prediction. This function numbers classes in sorted order, so the encoding depends only on the set of labels present, never on their order in the file.

Parameters

• labels - The per-sample labels to encode.

Returns

• (Array1<usize>, Vec<T>) - The per-sample codes, and the sorted class list where index i is the class encoded as i.

Errors

• DatasetError::ValidationError - Returns this when labels is empty.

Example

use dataset_ml::preprocessing::label_encode;
use ndarray::array;

let labels = array!["virginica", "setosa", "setosa", "versicolor"];
let (codes, classes) = label_encode(&labels).unwrap();

// Classes get numbers alphabetically, not in order of appearance.
assert_eq!(classes, vec!["setosa", "versicolor", "virginica"]);
assert_eq!(codes, array![2, 0, 0, 1]);

// Decode a prediction through the class list.
assert_eq!(classes[codes[0]], "virginica");