Enum SamplingStrategy 

#[non_exhaustive]
pub enum SamplingStrategy {
    Sequential,
    RandomShuffle,
    Stratified(Vec<usize>),
    WeightedRandom(Vec<f64>),
}

How rows are ordered / selected when building batches.

#[non_exhaustive] allows future strategies to be added without a breaking change.

Variants (Non-exhaustive)

Non-exhaustive enums could have additional variants added in future. Therefore, when matching against variants of non-exhaustive enums, an extra wildcard arm must be added to account for any future variants.

Sequential

Rows are yielded in their natural (insertion) order.

RandomShuffle

Rows are globally shuffled at the start of each epoch.

Stratified(Vec<usize>)

Stratified sampling: rows are first grouped by the provided class indices, then interleaved so every batch contains representatives of every class present in the dataset.

The inner Vec<usize> maps each dataset row index to its class label (integer encoded).

WeightedRandom(Vec<f64>)

Weighted random sampling without replacement (within an epoch).

The inner Vec<f64> provides a non-negative weight for every row. Rows with higher weights are more likely to appear early in the epoch ordering.

Trait Implementations

impl Clone for SamplingStrategy

fn clone(&self) -> SamplingStrategy

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SamplingStrategy

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

Formats the value using the given formatter.