Struct TimeSeriesSplit 

pub struct TimeSeriesSplit {
    pub n_splits: usize,
    pub min_train_size: Option<usize>,
    pub max_train_size: Option<usize>,
}

Time series split for temporal data.

Respects the temporal order of data. Each training set consists of data before the validation set (no data leakage from future).

Fields

n_splits: usize

Number of splits.

min_train_size: Option<usize>

Minimum training set size.

max_train_size: Option<usize>

Maximum training set size (for sliding window).

Implementations

impl TimeSeriesSplit

pub fn new(n_splits: usize) -> TrainResult<Self>

Create a new time series split.

Arguments

• n_splits - Number of splits (must be >= 2)

pub fn with_min_train_size(self, size: usize) -> Self

Set minimum training set size.

pub fn with_max_train_size(self, size: usize) -> Self

Set maximum training set size (for sliding window).

Trait Implementations

impl Clone for TimeSeriesSplit

fn clone(&self) -> TimeSeriesSplit

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl CrossValidationSplit for TimeSeriesSplit

fn num_splits(&self) -> usize

Get the number of splits.

fn get_split( &self, fold: usize, n_samples: usize, ) -> TrainResult<(Vec<usize>, Vec<usize>)>

Get the train/validation indices for a specific fold.

impl Debug for TimeSeriesSplit

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

Formats the value using the given formatter.