Skip to content

sklearn.md

Latest commit

 

History

History
535 lines (416 loc) · 30.5 KB

sklearn.md

File metadata and controls

535 lines (416 loc) · 30.5 KB

scikit-learn (sklearn) - Pure NumPy Implementation

Version: 1.8.0-offlinai | Type: Pure Python/NumPy | Size: 12,077 lines across 40 modules | Location: sklearn/

A from-scratch reimplementation of scikit-learn using only NumPy. No compiled extensions needed -- runs on iOS/iPadOS natively. Implements the full scikit-learn estimator API (fit, predict, transform, score, get_params, set_params).

Quick Start

from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score

X, y = make_classification(n_samples=200, n_features=4, random_state=42)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3)

clf = RandomForestClassifier(n_estimators=10, max_depth=5)
clf.fit(X_train, y_train)
print(f"Accuracy: {accuracy_score(y_test, clf.predict(X_test)):.3f}")

Module Reference

sklearn.linear_model -- Linear Models

Class Key Parameters Methods
LinearRegression fit_intercept fit(), predict(), score()
Ridge alpha, fit_intercept, solver fit(), predict(), score()
Lasso alpha, max_iter, tol fit(), predict(), score()
ElasticNet alpha, l1_ratio, max_iter, tol fit(), predict(), score()
LogisticRegression C, max_iter, lr, penalty, solver, multi_class fit(), predict(), predict_proba(), score(), decision_function()
SGDClassifier loss, alpha, max_iter, learning_rate, eta0, penalty fit(), partial_fit(), predict(), decision_function(), score()
SGDRegressor loss, alpha, max_iter, learning_rate, eta0, penalty fit(), partial_fit(), predict(), score()
RidgeClassifier alpha, fit_intercept fit(), predict(), decision_function(), score()
Perceptron alpha, max_iter, eta0, penalty fit(), predict(), score()
BayesianRidge max_iter, tol, alpha_1, alpha_2, lambda_1, lambda_2 fit(), predict(), score()
ARDRegression max_iter, tol, alpha_1, alpha_2, lambda_1, lambda_2, threshold_lambda fit(), predict(), score()
HuberRegressor epsilon, max_iter, alpha fit(), predict(), score()
Lars n_nonzero_coefs, fit_intercept fit(), predict(), score()
LassoLars alpha, max_iter, fit_intercept fit(), predict(), score()
OrthogonalMatchingPursuit n_nonzero_coefs, tol fit(), predict(), score()
PoissonRegressor alpha, max_iter, tol fit(), predict(), score()
GammaRegressor alpha, max_iter, tol fit(), predict(), score()
TweedieRegressor power, alpha, max_iter, tol fit(), predict(), score()

sklearn.tree -- Decision Trees

Class Key Parameters Methods
DecisionTreeClassifier max_depth, min_samples_split, min_samples_leaf, criterion, max_features, random_state fit(), predict(), predict_proba(), score(), feature_importances_
DecisionTreeRegressor max_depth, min_samples_split, min_samples_leaf, criterion, max_features, random_state fit(), predict(), score(), feature_importances_
ExtraTreeClassifier max_depth, min_samples_split, max_features, random_state fit(), predict(), predict_proba(), score()
ExtraTreeRegressor max_depth, min_samples_split, max_features, random_state fit(), predict(), score()

sklearn.ensemble -- Ensemble Methods

Class Key Parameters Methods
RandomForestClassifier n_estimators, max_depth, max_features, min_samples_split, random_state, n_jobs fit(), predict(), predict_proba(), score(), feature_importances_
RandomForestRegressor n_estimators, max_depth, max_features, min_samples_split, random_state fit(), predict(), score(), feature_importances_
GradientBoostingClassifier n_estimators, learning_rate, max_depth, subsample, loss, random_state fit(), predict(), predict_proba(), score(), feature_importances_
GradientBoostingRegressor n_estimators, learning_rate, max_depth, subsample, loss, random_state fit(), predict(), score(), feature_importances_
AdaBoostClassifier n_estimators, learning_rate, random_state fit(), predict(), predict_proba(), score()
AdaBoostRegressor n_estimators, learning_rate, loss, random_state fit(), predict(), score()
BaggingClassifier n_estimators, max_samples, max_features, random_state fit(), predict(), predict_proba(), score()
BaggingRegressor n_estimators, max_samples, max_features, random_state fit(), predict(), score()
ExtraTreesClassifier n_estimators, max_depth, max_features, random_state fit(), predict(), predict_proba(), score(), feature_importances_
ExtraTreesRegressor n_estimators, max_depth, max_features, random_state fit(), predict(), score(), feature_importances_
HistGradientBoostingClassifier max_iter, learning_rate, max_depth, max_leaf_nodes, min_samples_leaf, l2_regularization fit(), predict(), predict_proba(), score()
HistGradientBoostingRegressor max_iter, learning_rate, max_depth, max_leaf_nodes, min_samples_leaf, l2_regularization fit(), predict(), score()
IsolationForest n_estimators, max_samples, contamination, random_state fit(), predict(), decision_function(), score_samples()
VotingClassifier estimators, voting (hard/soft), weights fit(), predict(), predict_proba(), score()
VotingRegressor estimators, weights fit(), predict(), score()
StackingClassifier estimators, final_estimator, cv fit(), predict(), predict_proba(), score()
StackingRegressor estimators, final_estimator, cv fit(), predict(), score()

sklearn.cluster -- Clustering

Class Key Parameters Methods / Attributes
KMeans n_clusters, n_init, max_iter, random_state, init fit(), predict(), fit_predict(), labels_, cluster_centers_, inertia_
MiniBatchKMeans n_clusters, batch_size, max_iter, random_state fit(), predict(), fit_predict(), partial_fit(), labels_, cluster_centers_
DBSCAN eps, min_samples, metric fit(), fit_predict(), labels_, core_sample_indices_
AgglomerativeClustering n_clusters, linkage (single/complete/average/ward), distance_threshold fit(), fit_predict(), labels_, n_clusters_
SpectralClustering n_clusters, affinity, gamma, random_state fit(), fit_predict(), labels_
MeanShift bandwidth, seeds, bin_seeding fit(), predict(), fit_predict(), labels_, cluster_centers_
OPTICS min_samples, max_eps, metric, cluster_method fit(), labels_, reachability_, ordering_
Birch threshold, branching_factor, n_clusters fit(), predict(), fit_predict(), labels_
AffinityPropagation damping, max_iter, preference fit(), predict(), fit_predict(), labels_, cluster_centers_indices_
BisectingKMeans n_clusters, n_init, max_iter, random_state fit(), predict(), fit_predict(), labels_
HDBSCAN min_cluster_size, min_samples, cluster_selection_epsilon fit(), fit_predict(), labels_, probabilities_
FeatureAgglomeration n_clusters, linkage, distance_threshold fit(), transform(), fit_transform(), labels_

sklearn.preprocessing -- Data Preprocessing

Class Key Parameters Methods
StandardScaler with_mean, with_std fit(), transform(), fit_transform(), inverse_transform()
MinMaxScaler feature_range fit(), transform(), fit_transform(), inverse_transform()
RobustScaler with_centering, with_scaling, quantile_range fit(), transform(), fit_transform(), inverse_transform()
MaxAbsScaler -- fit(), transform(), fit_transform(), inverse_transform()
Normalizer norm (l1/l2/max) fit(), transform(), fit_transform()
Binarizer threshold fit(), transform(), fit_transform()
LabelEncoder -- fit(), transform(), fit_transform(), inverse_transform(), classes_
OneHotEncoder sparse_output, handle_unknown, drop fit(), transform(), fit_transform(), inverse_transform(), categories_
OrdinalEncoder categories, handle_unknown fit(), transform(), fit_transform(), inverse_transform()
LabelBinarizer neg_label, pos_label fit(), transform(), fit_transform(), inverse_transform()
PolynomialFeatures degree, interaction_only, include_bias fit(), transform(), fit_transform(), n_output_features_
PowerTransformer method (yeo-johnson/box-cox), standardize fit(), transform(), fit_transform(), inverse_transform()
QuantileTransformer n_quantiles, output_distribution (uniform/normal) fit(), transform(), fit_transform(), inverse_transform()
KBinsDiscretizer n_bins, encode (ordinal/onehot), strategy (uniform/quantile/kmeans) fit(), transform(), fit_transform()
FunctionTransformer func, inverse_func fit(), transform(), fit_transform(), inverse_transform()
SplineTransformer n_knots, degree, knots fit(), transform(), fit_transform()
TargetEncoder categories, smooth fit(), transform(), fit_transform()

sklearn.decomposition -- Dimensionality Reduction

Class Key Parameters Methods / Attributes
PCA n_components, whiten, svd_solver fit(), transform(), fit_transform(), inverse_transform(), explained_variance_ratio_, components_
TruncatedSVD n_components, n_iter, random_state fit(), transform(), fit_transform(), explained_variance_ratio_, components_
NMF n_components, init, max_iter, tol, random_state fit(), transform(), fit_transform(), components_
FastICA n_components, algorithm, max_iter, tol, random_state fit(), transform(), fit_transform(), mixing_, components_
KernelPCA n_components, kernel, gamma, degree fit(), transform(), fit_transform()
IncrementalPCA n_components, batch_size fit(), transform(), partial_fit(), explained_variance_ratio_
LatentDirichletAllocation n_components, max_iter, learning_method, random_state fit(), transform(), fit_transform(), components_
SparsePCA n_components, alpha, max_iter, random_state fit(), transform(), fit_transform(), components_
FactorAnalysis n_components, max_iter, tol fit(), transform(), fit_transform(), components_
DictionaryLearning n_components, alpha, max_iter, random_state fit(), transform(), fit_transform(), components_
MiniBatchNMF n_components, batch_size, max_iter, random_state fit(), transform(), partial_fit(), components_

sklearn.neighbors -- Nearest Neighbors

Class Key Parameters Methods
KNeighborsClassifier n_neighbors, weights (uniform/distance), metric, p fit(), predict(), predict_proba(), score(), kneighbors()
KNeighborsRegressor n_neighbors, weights, metric, p fit(), predict(), score(), kneighbors()
RadiusNeighborsClassifier radius, weights, metric fit(), predict(), predict_proba(), score()
RadiusNeighborsRegressor radius, weights, metric fit(), predict(), score()
NearestNeighbors n_neighbors, radius, metric fit(), kneighbors(), radius_neighbors()
NearestCentroid metric, shrink_threshold fit(), predict(), score(), centroids_
LocalOutlierFactor n_neighbors, contamination, metric fit(), fit_predict(), decision_function(), negative_outlier_factor_
KernelDensity bandwidth, kernel, metric fit(), score_samples(), score(), sample()

sklearn.svm -- Support Vector Machines

Class Key Parameters Methods
SVC C, kernel, gamma, degree, max_iter fit(), predict(), decision_function(), score()
SVR C, kernel, gamma, epsilon, max_iter fit(), predict(), score()
LinearSVC C, loss, penalty, max_iter, dual fit(), predict(), decision_function(), score()
LinearSVR C, epsilon, loss, max_iter fit(), predict(), score()
NuSVC nu, kernel, gamma, degree, max_iter fit(), predict(), decision_function(), score()
NuSVR nu, C, kernel, gamma, max_iter fit(), predict(), score()
OneClassSVM nu, kernel, gamma, degree fit(), predict(), decision_function(), score_samples()

sklearn.naive_bayes -- Naive Bayes

Class Key Parameters Methods
GaussianNB var_smoothing, priors fit(), predict(), predict_proba(), predict_log_proba(), score(), partial_fit()
MultinomialNB alpha, fit_prior fit(), predict(), predict_proba(), predict_log_proba(), score(), partial_fit()
BernoulliNB alpha, binarize, fit_prior fit(), predict(), predict_proba(), predict_log_proba(), score(), partial_fit()
ComplementNB alpha, norm fit(), predict(), predict_proba(), score()
CategoricalNB alpha, fit_prior fit(), predict(), predict_proba(), score()

sklearn.neural_network -- Neural Networks

Class Key Parameters Methods
MLPClassifier hidden_layer_sizes, activation (relu/tanh/logistic), solver (adam/sgd), max_iter, learning_rate_init, alpha fit(), predict(), predict_proba(), score(), partial_fit()
MLPRegressor hidden_layer_sizes, activation, solver, max_iter, learning_rate_init, alpha fit(), predict(), score(), partial_fit()

sklearn.gaussian_process -- Gaussian Processes

Class Key Parameters Methods
GaussianProcessClassifier kernel, n_restarts_optimizer, random_state fit(), predict(), predict_proba(), score(), log_marginal_likelihood()
GaussianProcessRegressor kernel, alpha, n_restarts_optimizer, random_state fit(), predict(), score(), log_marginal_likelihood()

Kernels: RBF, Matern, RationalQuadratic, ExpSineSquared, DotProduct, WhiteKernel, ConstantKernel, Sum, Product

sklearn.discriminant_analysis -- Discriminant Analysis

Class Key Parameters Methods
LinearDiscriminantAnalysis n_components, solver, shrinkage fit(), predict(), predict_proba(), transform(), score()
QuadraticDiscriminantAnalysis reg_param fit(), predict(), predict_proba(), score()

sklearn.mixture -- Gaussian Mixture Models

Class Key Parameters Methods
GaussianMixture n_components, covariance_type (full/tied/diag/spherical), max_iter, random_state fit(), predict(), predict_proba(), score(), score_samples(), sample(), bic(), aic()
BayesianGaussianMixture n_components, covariance_type, weight_concentration_prior_type, max_iter fit(), predict(), predict_proba(), score(), score_samples()

sklearn.model_selection -- Model Selection & Validation

Function / Class Description
train_test_split(X, y, test_size, random_state, stratify) Split arrays into train/test subsets
cross_val_score(estimator, X, y, cv, scoring) Evaluate estimator by cross-validation
cross_validate(estimator, X, y, cv, scoring, return_train_score) Evaluate with multiple metrics
learning_curve(estimator, X, y, train_sizes, cv) Generate learning curve data
validation_curve(estimator, X, y, param_name, param_range, cv) Generate validation curve data
KFold(n_splits, shuffle, random_state) K-fold cross-validation splitter
StratifiedKFold(n_splits, shuffle, random_state) Stratified K-fold (preserves class ratios)
LeaveOneOut() Leave-one-out cross-validation
TimeSeriesSplit(n_splits, max_train_size) Time-series-aware cross-validation
ShuffleSplit(n_splits, test_size, random_state) Random permutation cross-validation
RepeatedKFold(n_splits, n_repeats, random_state) Repeated K-fold
RepeatedStratifiedKFold(n_splits, n_repeats, random_state) Repeated stratified K-fold
GroupKFold(n_splits) K-fold with non-overlapping groups
GridSearchCV(estimator, param_grid, cv, scoring, refit) Exhaustive grid search over parameters
RandomizedSearchCV(estimator, param_distributions, n_iter, cv, scoring) Randomized parameter search

sklearn.metrics -- Evaluation Metrics (38 functions)

Classification Metrics

Function Description
accuracy_score(y_true, y_pred) Fraction of correct predictions
balanced_accuracy_score(y_true, y_pred) Balanced accuracy (macro recall)
precision_score(y_true, y_pred, average) Precision (positive predictive value)
recall_score(y_true, y_pred, average) Recall (sensitivity / true positive rate)
f1_score(y_true, y_pred, average) F1 score (harmonic mean of precision/recall)
fbeta_score(y_true, y_pred, beta, average) F-beta score with adjustable beta
confusion_matrix(y_true, y_pred) Confusion matrix (NxN array)
classification_report(y_true, y_pred) Text report with precision/recall/F1 per class
roc_auc_score(y_true, y_score) Area under ROC curve
roc_curve(y_true, y_score) ROC curve (fpr, tpr, thresholds)
precision_recall_curve(y_true, probas_pred) Precision-recall curve
average_precision_score(y_true, y_score) Average precision (area under PR curve)
log_loss(y_true, y_pred) Log loss / cross-entropy
brier_score_loss(y_true, y_prob) Brier score for probabilistic predictions
cohen_kappa_score(y1, y2) Cohen's kappa inter-rater agreement
matthews_corrcoef(y_true, y_pred) Matthews correlation coefficient
hamming_loss(y_true, y_pred) Hamming loss (fraction of wrong labels)
jaccard_score(y_true, y_pred, average) Jaccard similarity coefficient
zero_one_loss(y_true, y_pred) Zero-one loss (fraction incorrect)
hinge_loss(y_true, pred_decision) Average hinge loss
top_k_accuracy_score(y_true, y_score, k) Top-k accuracy

Regression Metrics

Function Description
mean_squared_error(y_true, y_pred) Mean squared error
mean_absolute_error(y_true, y_pred) Mean absolute error
root_mean_squared_error(y_true, y_pred) Root mean squared error
median_absolute_error(y_true, y_pred) Median absolute error
r2_score(y_true, y_pred) R-squared (coefficient of determination)
explained_variance_score(y_true, y_pred) Explained variance
max_error(y_true, y_pred) Maximum residual error
mean_absolute_percentage_error(y_true, y_pred) MAPE
mean_squared_log_error(y_true, y_pred) MSLE
mean_pinball_loss(y_true, y_pred, alpha) Pinball (quantile) loss
d2_pinball_score(y_true, y_pred, alpha) D2 pinball score

Clustering Metrics

Function Description
silhouette_score(X, labels) Mean silhouette coefficient
silhouette_samples(X, labels) Per-sample silhouette values
calinski_harabasz_score(X, labels) Variance ratio criterion
davies_bouldin_score(X, labels) Davies-Bouldin index
adjusted_rand_score(labels_true, labels_pred) Adjusted Rand index
adjusted_mutual_info_score(labels_true, labels_pred) Adjusted mutual information
normalized_mutual_info_score(labels_true, labels_pred) Normalized mutual information

Pairwise & Distance

Function Description
pairwise_distances(X, Y, metric) Pairwise distance matrix
euclidean_distances(X, Y) Euclidean distance matrix
cosine_similarity(X, Y) Cosine similarity matrix

sklearn.datasets -- Dataset Generators & Loaders

Function Description
make_classification(n_samples, n_features, n_informative, n_classes, random_state) Synthetic classification dataset
make_regression(n_samples, n_features, n_informative, noise, random_state) Synthetic regression dataset
make_blobs(n_samples, centers, n_features, cluster_std, random_state) Gaussian blob clusters
make_moons(n_samples, noise, random_state) Two interleaving half-circles
make_circles(n_samples, noise, factor, random_state) Concentric circles
make_swiss_roll(n_samples, noise, random_state) 3D Swiss roll manifold
make_s_curve(n_samples, noise, random_state) 3D S-curve manifold
make_friedman1(n_samples, n_features, noise, random_state) Friedman #1 regression
make_friedman2(n_samples, noise, random_state) Friedman #2 regression
make_friedman3(n_samples, noise, random_state) Friedman #3 regression
load_iris() 150 samples, 4 features, 3 classes (setosa/versicolor/virginica)
load_digits() 1,797 samples, 64 features (8x8 pixel images), 10 classes
load_wine() 178 samples, 13 features, 3 classes (wine cultivars)
load_breast_cancer() 569 samples, 30 features, 2 classes (malignant/benign)
load_diabetes() 442 samples, 10 features (regression target: disease progression)

All loaders return a dict with keys: data, target, feature_names, target_names, DESCR.

sklearn.manifold -- Manifold Learning

Class Key Parameters Methods
TSNE n_components, perplexity, learning_rate, n_iter, random_state, metric fit_transform(), embedding_
MDS n_components, metric, max_iter, random_state fit(), fit_transform(), embedding_, stress_
Isomap n_components, n_neighbors fit(), transform(), fit_transform()
LocallyLinearEmbedding n_components, n_neighbors, method (standard/modified/hessian/ltsa) fit(), transform(), fit_transform()
SpectralEmbedding n_components, affinity, gamma, random_state fit(), fit_transform(), embedding_

sklearn.pipeline -- Pipeline Utilities

Class / Function Description
Pipeline(steps) Chain of transforms with a final estimator. Supports fit(), predict(), transform(), score(), set_params(), get_params()
make_pipeline(*steps) Convenience constructor for Pipeline (auto-generates step names)
FeatureUnion(transformer_list) Concatenate results of multiple transformers

sklearn.compose -- Column Transformers

Class Description
ColumnTransformer(transformers, remainder) Apply different transformers to different columns. transformers is a list of (name, transformer, columns) tuples. remainder='drop' or 'passthrough'
make_column_transformer(*transformers) Convenience constructor for ColumnTransformer
make_column_selector(pattern, dtype_include, dtype_exclude) Column selector for ColumnTransformer
TransformedTargetRegressor(regressor, transformer) Transform target before fitting a regressor

sklearn.impute -- Missing Value Imputation

Class Key Parameters Methods
SimpleImputer strategy (mean/median/most_frequent/constant), fill_value fit(), transform(), fit_transform()
KNNImputer n_neighbors, weights, metric fit(), transform(), fit_transform()
IterativeImputer estimator, max_iter, random_state fit(), transform(), fit_transform()
MissingIndicator features (missing-only/all) fit(), transform(), fit_transform()

sklearn.feature_extraction -- Feature Extraction

Class Description
DictVectorizer(sparse) Convert list of dicts to feature matrix
text.CountVectorizer(max_features, stop_words, ngram_range) Convert text to token count matrix
text.TfidfVectorizer(max_features, stop_words, ngram_range) Convert text to TF-IDF matrix
text.TfidfTransformer(norm, use_idf) Transform count matrix to TF-IDF
text.HashingVectorizer(n_features) Hash-based text vectorization

sklearn.feature_selection -- Feature Selection

Class Description
SelectKBest(score_func, k) Select K highest-scoring features
SelectPercentile(score_func, percentile) Select features by percentile
VarianceThreshold(threshold) Remove low-variance features
RFE(estimator, n_features_to_select) Recursive feature elimination
RFECV(estimator, cv) RFE with cross-validated feature count
SelectFromModel(estimator, threshold) Select features from model importance
SequentialFeatureSelector(estimator, n_features_to_select, direction) Forward/backward feature selection

Score functions: f_classif, f_regression, chi2, mutual_info_classif, mutual_info_regression

sklearn.calibration -- Probability Calibration

Class Key Parameters Methods
CalibratedClassifierCV estimator, method (sigmoid/isotonic), cv fit(), predict(), predict_proba(), score()
Function Description
calibration_curve(y_true, y_prob, n_bins) Compute true and predicted probabilities for calibration plot

sklearn.inspection -- Model Inspection

Function Description
permutation_importance(estimator, X, y, n_repeats, scoring) Permutation-based feature importance
partial_dependence(estimator, X, features) Partial dependence of features
PartialDependenceDisplay.from_estimator(estimator, X, features) Plot partial dependence

sklearn.multiclass -- Multiclass Strategies

Class Description
OneVsRestClassifier(estimator) One-vs-rest (OVR) multiclass strategy
OneVsOneClassifier(estimator) One-vs-one (OVO) multiclass strategy
OutputCodeClassifier(estimator, code_size) Error-correcting output codes

sklearn.multioutput -- Multi-Output

Class Description
MultiOutputClassifier(estimator) Multi-target classification
MultiOutputRegressor(estimator) Multi-target regression
ClassifierChain(base_estimator, order) Chained multi-label classification
RegressorChain(base_estimator, order) Chained multi-target regression

sklearn.semi_supervised -- Semi-Supervised Learning

Class Key Parameters Methods
LabelPropagation kernel, gamma, n_neighbors, max_iter fit(), predict(), predict_proba(), score()
LabelSpreading kernel, gamma, alpha, max_iter fit(), predict(), predict_proba(), score()
SelfTrainingClassifier base_estimator, threshold, max_iter fit(), predict(), predict_proba(), score()

sklearn.isotonic -- Isotonic Regression

Class Key Parameters Methods
IsotonicRegression increasing, out_of_bounds fit(), predict(), transform(), score()

sklearn.kernel_approximation -- Kernel Approximation

Class Description
RBFSampler(gamma, n_components) Approximate RBF kernel via random Fourier features
Nystroem(kernel, gamma, n_components) Approximate kernel map using Nystroem method
AdditiveChi2Sampler(sample_steps) Approximate additive chi-squared kernel

sklearn.cross_decomposition -- Cross Decomposition

Class Description
PLSRegression(n_components) Partial Least Squares regression
PLSCanonical(n_components) Canonical PLS
CCA(n_components) Canonical Correlation Analysis

sklearn.base -- Base Classes & Utilities

Function / Class Description
BaseEstimator Base class with get_params() / set_params()
ClassifierMixin Mixin adding score() for classifiers
RegressorMixin Mixin adding score() for regressors
TransformerMixin Mixin adding fit_transform()
ClusterMixin Mixin adding fit_predict()
clone(estimator) Create a new estimator with same parameters

Example: Full Pipeline

from sklearn.pipeline import Pipeline
from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.impute import SimpleImputer
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.model_selection import cross_val_score
from sklearn.datasets import make_classification

X, y = make_classification(n_samples=500, n_features=10, random_state=42)
pipe = Pipeline([
    ('scaler', StandardScaler()),
    ('clf', GradientBoostingClassifier(n_estimators=50, max_depth=3))
])
scores = cross_val_score(pipe, X, y, cv=5)
print(f"CV Accuracy: {scores.mean():.3f} +/- {scores.std():.3f}")

Compatibility Notes

  • All estimators follow the scikit-learn API: fit(), predict(), transform(), score(), get_params(), set_params()
  • NumPy arrays are used throughout (no pandas dependency)
  • Random states are reproducible via random_state parameter
  • No sparse matrix support (all operations use dense arrays)
  • No joblib parallelism (n_jobs parameter is accepted but ignored)