pub fn online_evaluate(
learner: &mut dyn OnlineLearner,
data: &[(Vec<f64>, f64)],
train: bool,
) -> Result<(Vec<f64>, OnlineStats), OnlineError>Expand description
Evaluate an online learner sequentially on a dataset.
When train = true, each sample is used to update the model (prequential
evaluation: predict first, then learn). When train = false, only
predictions are made and the model is not updated.
Returns (predictions, stats) where predictions[i] is the prediction
made before any update on sample i.