Expand description
Pure statistical computation — no plotters dependency.
Everything a chart type needs to know what to draw (quartiles, KDE curves,
ROC/PR points, AUC) is computed here and unit-tested against hand-computed
reference values, independently of any rendering. The series
layer consumes these outputs and turns them into draw calls.
Re-exports§
pub use calibration::calibration_curve;pub use calibration::CalibrationBin;pub use correlation::correlation_matrix;pub use correlation::pearson;pub use correlation::rank;pub use correlation::spearman;pub use correlation::CorrelationMethod;pub use ecdf::dkw_epsilon;pub use ecdf::ecdf;pub use ecdf::Ecdf as EcdfData;pub use gain::gain_curve;pub use gain::GainPoint;pub use histogram::histogram;pub use histogram::BinRule;pub use histogram::Histogram;pub use kde::kde_curve;pub use kde::silverman_bandwidth;pub use kde::KdeCurve;pub use normal::norm_ppf;pub use precision_recall::precision_recall_curve;pub use precision_recall::PrCurve;pub use precision_recall::PrPoint;pub use quartiles::quartiles;pub use quartiles::Quartiles;pub use roc::roc_curve;pub use roc::RocCurveData;pub use roc::RocPoint;
Modules§
- calibration
- Calibration (reliability) binning for probabilistic classifiers.
- correlation
- Pearson and Spearman correlation, and correlation matrices.
- ecdf
- Empirical cumulative distribution function (ECDF) and its DKW confidence band.
- gain
- Cumulative gain and lift curves for ranking/classification evaluation.
- histogram
- Histogram binning with several standard bin-count rules.
- kde
- Gaussian kernel density estimation, used to shape violin plots.
- normal
- Standard-normal inverse CDF (quantile function), for theoretical quantiles on Q–Q plots.
- precision_
recall - Precision–recall curve: threshold sweep producing (recall, precision) points and average precision (AP).
- quartiles
- Quartiles, IQR, and Tukey whisker/outlier bounds for box plots.
- roc
- ROC curve: threshold sweep producing (FPR, TPR) points, and AUC via trapezoidal integration.
Enums§
- Stats
Error - Errors returned by the statistics routines for inputs where a metric is
undefined, rather than letting
NaNpropagate silently into a chart.