100% Coverage Plan: sklearn-java

Current Status: ~40/400+ (~10%) — 19 submodules, 95 tests

Phase A — Core ML (High Impact, Most-Used APIs)

A1. Metrics + Model Selection

• Metrics: mean_absolute_error, mean_squared_error, root_mean_squared_error, r2_score, mean_absolute_percentage_error, median_absolute_error, max_error, explained_variance_score — all regression metrics
• roc_curve, roc_auc_score, auc, average_precision_score, precision_recall_curve, det_curve — ranking/classification metrics
• pairwise_distances, euclidean_distances, pairwise_kernels, nan_euclidean_distances — pairwise metrics
• silhouette_score, silhouette_samples, calinski_harabasz_score, davies_bouldin_score, adjusted_rand_score, mutual_info_score, adjusted_mutual_info_score, normalized_mutual_info_score, homogeneity_score, completeness_score, v_measure_score, fowlkes_mallows_score, rand_score — clustering metrics
• log_loss, hinge_loss, brier_score_loss, cohen_kappa_score, matthews_corrcoef, balanced_accuracy_score, hamming_loss, zero_one_loss, top_k_accuracy_score, jaccard_score, fbeta_score, classification_report — classification metrics
• Model Selection: KFold, StratifiedKFold, GroupKFold, LeaveOneOut, ShuffleSplit, StratifiedShuffleSplit, TimeSeriesSplit, RepeatedKFold, RepeatedStratifiedKFold, GroupShuffleSplit, PredefinedSplit, LeaveOneGroupOut, LeavePGroupsOut, LeavePOut
• cross_val_score, cross_validate, cross_val_predict
• GridSearchCV, RandomizedSearchCV, ParameterGrid, ParameterSampler
• learning_curve, validation_curve, permutation_test_score
• check_cv, make_scorer, get_scorer

A2. Ensemble (Remaining) + Tree Extras

• GradientBoostingClassifier, GradientBoostingRegressor
• BaggingClassifier, BaggingRegressor
• VotingClassifier, VotingRegressor
• StackingClassifier, StackingRegressor
• IsolationForest, HistGradientBoostingClassifier, HistGradientBoostingRegressor, RandomTreesEmbedding
• ExtraTreeClassifier, ExtraTreeRegressor (standalone)

A3. Linear Model (Remaining)

• SGDClassifier, SGDRegressor, SGDOneClassSVM
• RidgeCV, RidgeClassifier, RidgeClassifierCV
• LassoCV, LassoLars, LassoLarsCV, LassoLarsIC
• ElasticNetCV, MultiTaskElasticNet, MultiTaskElasticNetCV, MultiTaskLasso, MultiTaskLassoCV
• BayesianRidge, ARDRegression
• HuberRegressor, RANSACRegressor, TheilSenRegressor, QuantileRegressor
• Perceptron, PassiveAggressiveClassifier, PassiveAggressiveRegressor
• PoissonRegressor, GammaRegressor, TweedieRegressor
• OrthogonalMatchingPursuit, OrthogonalMatchingPursuitCV
• LogisticRegressionCV, Lars, LarsCV
• Functions: enet_path, lars_path, lasso_path, ridge_regression

A4. Naive Bayes + Neighbors (Remaining)

• MultinomialNB, BernoulliNB, ComplementNB, CategoricalNB
• NearestNeighbors, RadiusNeighborsClassifier, RadiusNeighborsRegressor
• LocalOutlierFactor, NearestCentroid, KernelDensity
• NeighborhoodComponentsAnalysis, KNeighborsTransformer, RadiusNeighborsTransformer
• Functions: kneighbors_graph, radius_neighbors_graph

A5. Neural Network + Feature Selection

• MLPClassifier, MLPRegressor, BernoulliRBM
• SelectKBest, SelectPercentile, SelectFpr, SelectFdr, SelectFwe, GenericUnivariateSelect
• RFE, RFECV, SelectFromModel, SequentialFeatureSelector
• Functions: chi2, f_classif, f_regression, mutual_info_classif, mutual_info_regression

Phase B — ML Workbench (Medium Impact)

B1. Decomposition (Remaining)

• NMF, MiniBatchNMF, FastICA, TruncatedSVD, KernelPCA, IncrementalPCA, FactorAnalysis
• DictionaryLearning, MiniBatchDictionaryLearning, SparsePCA, MiniBatchSparsePCA, SparseCoder
• LatentDirichletAllocation
• Functions: fastica, dict_learning, non_negative_factorization, randomized_svd

B2. Clustering (Remaining)

• AgglomerativeClustering, FeatureAgglomeration, Birch, OPTICS, SpectralClustering
• MeanShift, AffinityPropagation, MiniBatchKMeans, BisectingKMeans, HDBSCAN
• SpectralBiclustering, SpectralCoclustering
• Functions: k_means, mean_shift, affinity_propagation, spectral_clustering, dbscan, estimate_bandwidth, ward_tree, linkage_tree

B3. Impute + Datasets

• SimpleImputer, KNNImputer, IterativeImputer
• load_iris, load_breast_cancer, load_digits, load_diabetes, load_wine, load_linnerud
• make_classification, make_regression, make_blobs, make_moons, make_circles, make_friedman1, make_low_rank_matrix, make_spd_matrix, make_swiss_roll, make_s_curve, make_gaussian_quantiles, make_multilabel_classification, make_biclusters, make_checkerboard, make_sparse_spd_matrix, make_sparse_coded_signal, make_sparse_uncorrelated, make_hastie_10_2

B4. SVM Remaining + Tree Extras

• NuSVC, NuSVR, LinearSVC, LinearSVR, OneClassSVM
• l1_min_c function
• ExtraTreeClassifier, ExtraTreeRegressor (standalone from Ensemble)

Phase C — Specialized Modules (Niche Use Cases)

C1. Manifold Learning

• TSNE, Isomap, MDS, ClassicalMDS, SpectralEmbedding, LocallyLinearEmbedding
• Functions: locally_linear_embedding, smacof, spectral_embedding, trustworthiness

C2. Gaussian Process, Mixture, Cross Decomposition, LDA/QDA

• GaussianProcessRegressor, GaussianProcessClassifier, kernels
• GaussianMixture, BayesianGaussianMixture
• PLSRegression, PLSCanonical, CCA, PLSSVD
• LinearDiscriminantAnalysis, QuadraticDiscriminantAnalysis

C3. Multiclass, MultiOutput, Kernel Methods, Semi-Supervised

• OneVsRestClassifier, OneVsOneClassifier, OutputCodeClassifier
• MultiOutputRegressor, MultiOutputClassifier, ClassifierChain, RegressorChain
• KernelRidge, RBFSampler, SkewedChi2Sampler, AdditiveChi2Sampler, Nystroem, PolynomialCountSketch
• LabelPropagation, LabelSpreading, SelfTrainingClassifier

C4. Covariance, Feature Extraction, Isotonic, Pipeline Extras

• EmpiricalCovariance, LedoitWolf, ShrunkCovariance, MinCovDet, GraphicalLasso, GraphicalLassoCV, OAS, EllipticEnvelope
• DictVectorizer, FeatureHasher, CountVectorizer, TfidfVectorizer, TfidfTransformer
• IsotonicRegression
• FeatureUnion, ColumnTransformer

Execution Strategy

• Each phase → feature branch → PR → squash-merge to develop
• Each algorithm = class + JUnit 5 tests + checkstyle clean
• Deterministic: same seed → same results
• Javadoc on every public class
• Pipeline compatibility from day one