Module scoring 

Scoring: a lightweight metric abstraction usable standalone or, optionally, backed by smartcore::metrics.

The Scorer trait is deliberately tiny — a single score call over y_true / y_pred. Built-in scorers (accuracy, MAE, MSE, RMSE, R²) are implemented directly here so the crate needs no modeling dependency. For metrics that already exist in mature form elsewhere (F1, ROC-AUC, …), enable the smartcore-metrics feature to get smartcore_adapter rather than reimplementing them.

User-defined metrics are first-class: wrap any closure with make_scorer.

Modules

smartcore_adapter

Feature-gated adapters wrapping smartcore::metrics as Scorers.

Structs

Accuracy

Classification accuracy: the fraction of predictions that exactly match.

ClosureScorer

A Scorer backed by an arbitrary closure — the analogue of scikit-learn’s make_scorer, so users are never limited to the built-ins.

MeanAbsoluteError

Mean absolute error.

MeanSquaredError

Mean squared error.

R2Score

Coefficient of determination, R².

RootMeanSquaredError

Root mean squared error.

Traits

Scorer

A scoring metric over true and predicted target vectors.

Functions

make_scorer

Wrap a closure as a Scorer.