Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Crate reference

Solow is a Cargo workspace of 54 focused crates. Depend on the ones you need, or pull in the umbrella solow crate to re-export the full public API. This page lists the workspace.

Foundation

CratePurpose
solow-coreError types, numeric aliases, shared data-handling tools
solow-linalgPure-Rust linear algebra (SVD, eigh, Cholesky, QR, LU, pinv)
solow-distributionsSpecial functions, distributions, ECDF, GEV, GPD
solow-optimizeNewton / BFGS optimizers and numerical differentiation

Classical statistical models

CrateModels
solow-regressionOLS, WLS, GLS, GLSAR, penalised (Ridge, Lasso, ElasticNet), CV variants, quantile, rolling/recursive, LARS, LassoLars, LassoLarsIC, OMP, HuberRegressor, BayesianRidge, ARDRegression, RANSACRegressor, TheilSenRegressor, MultiTaskLasso/ElasticNet, SGD (regressor + classifier), Perceptron, PassiveAggressive*, RidgeClassifier(CV), Dummy*, KernelRidge, sliced inverse regression, robust covariances
solow-glmGeneralized linear models (families, links, IRLS) + Tweedie, PoissonRegressor, GammaRegressor, TweedieRegressor
solow-discreteLogit, Probit, Poisson, MNLogit, NegativeBinomial, GeneralizedPoisson, OrderedModel, ZeroInflatedPoisson, ConditionalLogit/Poisson, TruncatedLFPoisson, HurdleCountModel, LogisticRegressionCV
solow-robustRobust linear models (M-estimation)
solow-nonparametricNonparametric smoothers (lowess, KDE, kernel regression)
solow-multivariatePCA, factor analysis and rotation, MANOVA, canonical correlation
solow-durationSurvival analysis (Kaplan-Meier, Cox PH, log-rank)
solow-mixedLinear mixed-effects models (REML)
solow-geeGeneralized estimating equations (incl. nominal/ordinal)
solow-gamGeneralized additive models (penalized splines)
solow-imputeMICE combine + SimpleImputer, KnnImputer, IterativeImputer
solow-regimeRegime-switching models (Markov switching)
solow-othermodBeta regression
solow-copulaCopulas (Archimedean and elliptical)
solow-bayesBayesian mixed GLM (variational Bayes)
solow-emplikeEmpirical-likelihood inference

Time series

CrateContents
solow-tsaacf/pacf, ADF, KPSS, coint, Granger, AutoReg, ARMA, STL, seasonal_decompose, Holt-Winters, HP/BK/CF filters, zivot_andrews, range_unit_root, breakvar heteroskedasticity, GARCH(1,1), change-point detection (CUSUM, PELT, Binary Segmentation), EWMA control chart
solow-statespaceKalman filter/smoother, SARIMAX, unobserved components, dynamic factor
solow-varVAR, SVAR, VECM / Johansen

Statistics, formulas, and presentation

CratePurpose
solow-statsANOVA, Tukey HSD, ARCH, RESET, robust sandwich covariances (HC0-3, HAC, cluster), VIF, mediation, Oaxaca, power, proportion, rates, inter-rater (Cohen's / Fleiss' κ), influence, distance correlation, TOST equivalence, contingency, Pearson / Spearman / Kendall correlations, Mann-Whitney U, Kruskal-Wallis, McNemar, χ² contingency, Levene / Bartlett / Fligner variance tests, Shapiro-Wilk, Anderson-Darling, Kolmogorov-Smirnov, Wald-Wolfowitz runs, meta-analysis (fixed + DerSimonian-Laird random)
solow-formulaR/patsy-style formula interface (design matrices)
solow-summaryLabeled results and summary tables
solow-graphicsStatistical graphics (qqplot, plot_acf, influence)
solow-vizGeneral-purpose data-visualization backend (SVG); ROC / PR / reliability / residuals diagnostics

Model evaluation & resampling

CratePurpose
solow-metricsRegression / classification / forecast losses, calibration + post-hoc calibrators (Platt, isotonic, temperature), conformal (split, jackknife+), Diebold-Mariano and Giacomini-White, Friedman / Nemenyi / Wilcoxon, WAIC / PSIS-LOO, permutation importance, partial dependence, ALE, pairwise distances + kernels (RBF / linear / polynomial / sigmoid / laplacian / cosine / χ²), cluster evaluation (silhouette, ARI, MI variants, homogeneity / completeness / v-measure, Fowlkes-Mallows, Calinski-Harabasz, Davies-Bouldin), effect sizes (Cohen's d, Hedges' g, Glass's Δ, η², ω², Cliff's δ, Cramér's V), classification_report
solow-cvK-fold, stratified, walk-forward, leave-one-out, leave-p-out, shuffle-split, group-aware, purged / combinatorial-purged K-fold; repeated K-fold / stratified K-fold; stratified & group shuffle split; cross_val_score scoring; learning_curve, validation_curve, permutation_test_score; bootstrap CIs (percentile, basic, BCa); moving / circular / stationary block bootstrap for time series; optional rayon-parallel fold evaluation

Machine learning

CratePurpose
solow-preprocessing18 preprocessors: Standard/MinMax/MaxAbs/Robust scalers, Normalizer, LabelEncoder, OrdinalEncoder, OneHotEncoder, PolynomialFeatures, KBinsDiscretizer (uniform / quantile / KMeans), PowerTransformer (Yeo-Johnson / Box-Cox), QuantileTransformer, Binarizer, LabelBinarizer, MultiLabelBinarizer, FunctionTransformer, SplineTransformer, TargetEncoder
solow-cluster11 algorithms: KMeans (+K-means++), MiniBatchKMeans, BisectingKMeans, DBSCAN, HDBSCAN, OPTICS, MeanShift, AffinityPropagation, SpectralClustering, AgglomerativeClustering (single/complete/average/Ward), Birch + 2 mixtures: GaussianMixture, BayesianGaussianMixture
solow-neighborsKdTree, BallTree, KNeighborsClassifier / KNeighborsRegressor (uniform / distance weights), RadiusNeighborsClassifier / RadiusNeighborsRegressor, NearestCentroid, LocalOutlierFactor, KernelDensity (multivariate)
solow-treeDecisionTreeClassifier, DecisionTreeRegressor, ExtraTreeClassifier, ExtraTreeRegressor
solow-ensembleRandomForest*, ExtraTrees*, GradientBoosting*, HistGradientBoosting*, Bagging*, AdaBoost (classifier + regressor), Voting (classifier + regressor), Stacking (classifier + regressor), IsolationForest
solow-naive-bayesGaussianNB, MultinomialNB, BernoulliNB, ComplementNB, CategoricalNB
solow-discriminantLinearDiscriminantAnalysis, QuadraticDiscriminantAnalysis
solow-feature-selectionSelectKBest, RFE, VarianceThreshold, SelectPercentile, SelectFpr, SelectFdr, SelectFwe, SequentialFeatureSelector
solow-pipelinePipeline, FeatureUnion, ColumnTransformer, TransformedTargetRegressor, GridSearchCV, RandomizedSearchCV, HalvingGridSearchCV, HalvingRandomSearchCV
solow-svmLinearSVC, LinearSVR, SVC, SVR, NuSVC, NuSVR, OneClassSVM (kernels: linear / RBF / polynomial / sigmoid)
solow-neuralMLPClassifier, MLPRegressor (SGD + Adam), BernoulliRBM
solow-manifoldIsomap, LocallyLinearEmbedding, MDS, SpectralEmbedding, t-SNE
solow-decompositionPca, KernelPca, FastIca, Nmf, MiniBatchNmf, TruncatedSVD, IncrementalPCA, SparsePCA, DictionaryLearning, MiniBatchDictionaryLearning, LatentDirichletAllocation, GaussianRandomProjection, SparseRandomProjection
solow-textCountVectorizer, TfidfVectorizer, HashingVectorizer, FeatureHasher, DictVectorizer
solow-covarianceEmpiricalCovariance, ShrunkCovariance, LedoitWolf, Oas, MinCovDet, GraphicalLasso, EllipticEnvelope
solow-cross-decompositionPLSRegression, PLSCanonical, PLSSVD, CCA
solow-kernel-approxRBFSampler, Nystroem, AdditiveChi2Sampler, SkewedChi2Sampler, PolynomialCountSketch
solow-gpGaussianProcessRegressor + GaussianProcessClassifier + kernel algebra (RBF, Matern, RationalQuadratic, DotProduct, Constant, White, Sum, Product, Exponentiation)
solow-semi-supervisedLabelPropagation, LabelSpreading, SelfTrainingClassifier
solow-multiOneVsRest, OneVsOne, OutputCode, MultiOutputRegressor, MultiOutputClassifier, ClassifierChain, RegressorChain
solow-calibrationCalibratedClassifierCV (Platt sigmoid + isotonic)
solow-datasetsmake_classification / regression / blobs / moons / circles / swiss_roll / low_rank_matrix, load_iris / wine / diabetes / breast_cancer, compute_class_weight, compute_sample_weight, resample helpers

Umbrella and tooling

CratePurpose
solowUmbrella crate re-exporting the full public API, with a prelude
solow-pyPyO3 bindings: import solow from Python
solow-polarsPolars DataFrame interop
solow-benchcriterion benchmark harness

Every deterministic estimator is cross-verified against committed golden reference fixtures. Closed-form solvers agree bit-wise to 1e-10. Iterative solvers agree on parameters to 1e-6 or on predictions to 5e-2 where reference solvers themselves disagree at that scale.