# NVIDIA cuML > NVIDIA cuML 2020-2026, NVIDIA Corporation ## Pages in this subsection - [API Reference](index.html.md): This is the class and function reference of cuML. Please refer to the - [cuml](cuml/index.html.md): | `set_global_output_type` | Set the global output type. | - [set_global_output_type](generated/cuml.set_global_output_type/index.html.md): Set the global output type. - [using_output_type](generated/cuml.using_output_type/index.html.md): Configure the output type within a context. - [cuml.accel](cuml.accel/index.html.md): | `install` | Enable `cuml.accel`. | - [install](generated/cuml.accel.install/index.html.md): Enable `cuml.accel`. - [enabled](generated/cuml.accel.enabled/index.html.md): Returns whether the accelerator is enabled. - [profile](generated/cuml.accel.profile/index.html.md): Profile a section of code. - [is_proxy](generated/cuml.accel.is_proxy/index.html.md): Check if an instance or class is a proxy object created by the accelerator. - [cuml.benchmark](cuml.benchmark/index.html.md): Wraps a cuML algorithm and (optionally) a cpu-based algorithm - [cuml.cluster](cuml.cluster/index.html.md): | `AgglomerativeClustering` | Agglomerative Clustering ... - [AgglomerativeClustering](generated/cuml.cluster.AgglomerativeClustering/index.html.md): Agglomerative Clustering - [DBSCAN](generated/cuml.cluster.DBSCAN/index.html.md): DBSCAN is a very powerful yet fast clustering technique that finds clusters - [KMeans](generated/cuml.cluster.KMeans/index.html.md): KMeans is a basic but powerful clustering method which is optimized via - [SpectralClustering](generated/cuml.cluster.SpectralClustering/index.html.md): Apply spectral clustering from the normalized Laplacian. - [spectral_clustering](generated/cuml.cluster.spectral_clustering/index.html.md): Apply clustering to a projection of the normalized Laplacian. - [HDBSCAN](generated/cuml.cluster.hdbscan.HDBSCAN/index.html.md): HDBSCAN Clustering - [all_points_membership_vectors](generated/cuml.cluster.hdbscan.all_points_membership_vectors/index.html.md): Predict soft cluster membership vectors for all points in the - [membership_vector](generated/cuml.cluster.hdbscan.membership_vector/index.html.md): Predict soft cluster membership. The result produces a vector - [approximate_predict](generated/cuml.cluster.hdbscan.approximate_predict/index.html.md): Predict the cluster label of new points. The returned labels - [cuml.compose](cuml.compose/index.html.md): | `ColumnTransformer` | Applies transformers to columns of an array or dataframe. ... - [ColumnTransformer](generated/cuml.compose.ColumnTransformer/index.html.md): Applies transformers to columns of an array or dataframe. - [make_column_transformer](generated/cuml.compose.make_column_transformer/index.html.md): Construct a ColumnTransformer from the given transformers. - [make_column_selector](generated/cuml.compose.make_column_selector/index.html.md): Create a callable to select columns to be used with - [cuml.covariance](cuml.covariance/index.html.md): | `EmpiricalCovariance` | Maximum likelihood covariance estimator. | - [EmpiricalCovariance](generated/cuml.covariance.EmpiricalCovariance/index.html.md): Maximum likelihood covariance estimator. - [LedoitWolf](generated/cuml.covariance.LedoitWolf/index.html.md): LedoitWolf Estimator for covariance matrix estimation. - [cuml.dask](cuml.dask/index.html.md): Multi-node, multi-GPU algorithms using Dask. - [DBSCAN](generated/cuml.dask.cluster.DBSCAN/index.html.md): Multi-Node Multi-GPU implementation of DBSCAN. - [KMeans](generated/cuml.dask.cluster.KMeans/index.html.md): Multi-Node Multi-GPU implementation of KMeans. - [PCA](generated/cuml.dask.decomposition.PCA/index.html.md): PCA (Principal Component Analysis) is a fundamental dimensionality - [TruncatedSVD](generated/cuml.dask.decomposition.TruncatedSVD/index.html.md): * **Parameters:** - [RandomForestClassifier](generated/cuml.dask.ensemble.RandomForestClassifier/index.html.md): Multi-GPU Random Forest classifier model which fits multiple decision tree - [RandomForestRegressor](generated/cuml.dask.ensemble.RandomForestRegressor/index.html.md): Multi-GPU Random Forest regressor model which fits multiple decision tree - [LinearRegression](generated/cuml.dask.linear_model.LinearRegression/index.html.md): LinearRegression is a simple machine learning model where the response y is - [Ridge](generated/cuml.dask.linear_model.Ridge/index.html.md): Ridge extends LinearRegression by providing L2 regularization on the - [Lasso](generated/cuml.dask.linear_model.Lasso/index.html.md): Lasso extends LinearRegression by providing L1 regularization on the - [ElasticNet](generated/cuml.dask.linear_model.ElasticNet/index.html.md): ElasticNet extends LinearRegression with combined L1 and L2 regularizations - [UMAP](generated/cuml.dask.manifold.UMAP/index.html.md): Uniform Manifold Approximation and Projection - [MultinomialNB](generated/cuml.dask.naive_bayes.MultinomialNB/index.html.md): Distributed Naive Bayes classifier for multinomial models - [NearestNeighbors](generated/cuml.dask.neighbors.NearestNeighbors/index.html.md): Multi-node Multi-GPU NearestNeighbors Model. - [KNeighborsClassifier](generated/cuml.dask.neighbors.KNeighborsClassifier/index.html.md): Multi-node Multi-GPU K-Nearest Neighbors Classifier Model. - [KNeighborsRegressor](generated/cuml.dask.neighbors.KNeighborsRegressor/index.html.md): Multi-node Multi-GPU K-Nearest Neighbors Regressor Model. - [LabelBinarizer](generated/cuml.dask.preprocessing.LabelBinarizer/index.html.md): A distributed version of LabelBinarizer for one-hot encoding - [OneHotEncoder](generated/cuml.dask.preprocessing.OneHotEncoder/index.html.md): Encode categorical features as a one-hot numeric array. - [TfidfTransformer](generated/cuml.dask.feature_extraction.text.TfidfTransformer/index.html.md): Distributed TF-IDF transformer - [make_blobs](generated/cuml.dask.datasets.make_blobs/index.html.md): Makes labeled Dask-Cupy arrays containing blobs - [make_classification](generated/cuml.dask.datasets.make_classification/index.html.md): Generate a random n-class classification problem. - [make_regression](generated/cuml.dask.datasets.make_regression/index.html.md): Generate a random regression problem. - [CD](generated/cuml.dask.solvers.CD/index.html.md): Multi-node Multi-GPU Coordinate Descent Solver. - [BaseEstimator](generated/cuml.dask.common.base.BaseEstimator/index.html.md): | `get_combined_model`() | Return single-GPU model for serialization | - [DelayedParallelFunc](generated/cuml.dask.common.base.DelayedParallelFunc/index.html.md): Main documentation page - [DelayedPredictionMixin](generated/cuml.dask.common.base.DelayedPredictionMixin/index.html.md): Main documentation page - [DelayedTransformMixin](generated/cuml.dask.common.base.DelayedTransformMixin/index.html.md): Main documentation page - [DelayedInverseTransformMixin](generated/cuml.dask.common.base.DelayedInverseTransformMixin/index.html.md): Main documentation page - [cuml.datasets](cuml.datasets/index.html.md): | `make_blobs` | Generate isotropic Gaussian blobs for clustering. | - [make_blobs](generated/cuml.datasets.make_blobs/index.html.md): Generate isotropic Gaussian blobs for clustering. - [make_classification](generated/cuml.datasets.make_classification/index.html.md): Generate a random n-class classification problem. - [make_regression](generated/cuml.datasets.make_regression/index.html.md): Generate a random regression problem. - [cuml.decomposition](cuml.decomposition/index.html.md): | `PCA` | PCA (Principal Component Analysis) is a fundamental dimensionality r... - [PCA](generated/cuml.decomposition.PCA/index.html.md): PCA (Principal Component Analysis) is a fundamental dimensionality - [IncrementalPCA](generated/cuml.decomposition.IncrementalPCA/index.html.md): Based on sklearn.decomposition.IncrementalPCA from scikit-learn 0.23.1 - [TruncatedSVD](generated/cuml.decomposition.TruncatedSVD/index.html.md): TruncatedSVD is used to compute the top K singular values and vectors of a - [cuml.ensemble](cuml.ensemble/index.html.md): | `IsolationForest` | GPU-accelerated Isolation Forest for anomaly detection. ... - [IsolationForest](generated/cuml.ensemble.IsolationForest/index.html.md): GPU-accelerated Isolation Forest for anomaly detection. - [RandomForestClassifier](generated/cuml.ensemble.RandomForestClassifier/index.html.md): Implements a Random Forest classifier model which fits multiple decision - [RandomForestRegressor](generated/cuml.ensemble.RandomForestRegressor/index.html.md): Implements a Random Forest regressor model which fits multiple decision - [cuml.explainer](cuml.explainer/index.html.md): | `KernelExplainer` | GPU accelerated of SHAP's kernel explainer. ... - [KernelExplainer](generated/cuml.explainer.KernelExplainer/index.html.md): GPU accelerated of SHAP’s kernel explainer. - [PermutationExplainer](generated/cuml.explainer.PermutationExplainer/index.html.md): GPU accelerated version of SHAP’s PermutationExplainer - [TreeExplainer](generated/cuml.explainer.TreeExplainer/index.html.md): Model explainer that calculates Shapley values for the predictions of - [cuml.feature_extraction](cuml.feature_extraction/index.html.md): | `CountVectorizer` | Convert a collection of text documents to a matrix of token counts. | - [CountVectorizer](generated/cuml.feature_extraction.text.CountVectorizer/index.html.md): Convert a collection of text documents to a matrix of token counts. - [HashingVectorizer](generated/cuml.feature_extraction.text.HashingVectorizer/index.html.md): Convert a collection of text documents to a matrix of token occurrences. - [TfidfTransformer](generated/cuml.feature_extraction.text.TfidfTransformer/index.html.md): Transform a count matrix to a normalized tf or tf-idf representation. - [TfidfVectorizer](generated/cuml.feature_extraction.text.TfidfVectorizer/index.html.md): Convert a collection of raw documents to a matrix of TF-IDF features. - [cuml.kernel_ridge](cuml.kernel_ridge/index.html.md): | `KernelRidge` | Kernel ridge regression (KRR) performs l2 regularised ridge regression using the... - [KernelRidge](generated/cuml.kernel_ridge.KernelRidge/index.html.md): Kernel ridge regression (KRR) performs l2 regularised ridge regression - [cuml.linear_model](cuml.linear_model/index.html.md): | `LinearRegression` | Ordinary least squares Linear Regression. ... - [LinearRegression](generated/cuml.linear_model.LinearRegression/index.html.md): Ordinary least squares Linear Regression. - [LogisticRegression](generated/cuml.linear_model.LogisticRegression/index.html.md): Logistic Regression classifier. - [Ridge](generated/cuml.linear_model.Ridge/index.html.md): Linear least squares with L2 regularization. - [Lasso](generated/cuml.linear_model.Lasso/index.html.md): Linear Model trained with L1 prior as regularizer (aka the Lasso). - [ElasticNet](generated/cuml.linear_model.ElasticNet/index.html.md): Linear regression with combined L1 and L2 priors as regularizer. - [MBSGDClassifier](generated/cuml.linear_model.MBSGDClassifier/index.html.md): Linear models (linear SVM, logistic regression, or linear regression) - [MBSGDRegressor](generated/cuml.linear_model.MBSGDRegressor/index.html.md): Linear model fitted by minimizing a regularized empirical loss with SGD. - [Lars](generated/cuml.linear_model.Lars/index.html.md): Least Angle Regression - [cuml.manifold](cuml.manifold/index.html.md): | `UMAP` | Uniform Manifold Approximation and Proj... - [UMAP](generated/cuml.manifold.UMAP/index.html.md): Uniform Manifold Approximation and Projection - [TSNE](generated/cuml.manifold.TSNE/index.html.md): t-SNE (T-Distributed Stochastic Neighbor Embedding) is an extremely - [SpectralEmbedding](generated/cuml.manifold.SpectralEmbedding/index.html.md): Spectral embedding for non-linear dimensionality reduction. - [spectral_embedding](generated/cuml.manifold.spectral_embedding/index.html.md): Project the sample on the first eigenvectors of the graph Laplacian. - [fuzzy_simplicial_set](generated/cuml.manifold.umap.fuzzy_simplicial_set/index.html.md): Given a set of data X, a neighborhood size, and a measure of distance - [simplicial_set_embedding](generated/cuml.manifold.umap.simplicial_set_embedding/index.html.md): Perform a fuzzy simplicial set embedding, using a specified - [find_ab_params](generated/cuml.manifold.umap.find_ab_params/index.html.md): Fit a & b parameters for UMAP. - [cuml.metrics](cuml.metrics/index.html.md): | `accuracy_score` | Accuracy classification score. ... - [accuracy_score](generated/cuml.metrics.accuracy_score/index.html.md): Accuracy classification score. - [confusion_matrix](generated/cuml.metrics.confusion_matrix/index.html.md): Compute confusion matrix to evaluate the accuracy of a classification. - [kl_divergence](generated/cuml.metrics.kl_divergence/index.html.md): Calculates the “Kullback-Leibler” Divergence - [log_loss](generated/cuml.metrics.log_loss/index.html.md): Log loss, aka logistic loss or cross-entropy loss. - [roc_auc_score](generated/cuml.metrics.roc_auc_score/index.html.md): Compute Area Under the Receiver Operating Characteristic Curve (ROC AUC) - [precision_recall_curve](generated/cuml.metrics.precision_recall_curve/index.html.md): Compute precision-recall pairs for different probability thresholds - [trustworthiness](generated/cuml.metrics.trustworthiness/index.html.md): Expresses to what extent the local structure is retained in embedding. - [mean_absolute_error](generated/cuml.metrics.mean_absolute_error/index.html.md): Mean absolute error regression loss - [mean_squared_error](generated/cuml.metrics.mean_squared_error/index.html.md): Mean squared error regression loss - [mean_squared_log_error](generated/cuml.metrics.mean_squared_log_error/index.html.md): Mean squared log error regression loss - [median_absolute_error](generated/cuml.metrics.median_absolute_error/index.html.md): Median absolute error regression loss. - [r2_score](generated/cuml.metrics.r2_score/index.html.md): $R^2$ (coefficient of determination) regression score function. - [adjusted_rand_score](generated/cuml.metrics.cluster.adjusted_rand_score/index.html.md): Adjusted_rand_score is a clustering similarity metric based on the Rand - [entropy](generated/cuml.metrics.cluster.entropy/index.html.md): cython_entropy(clustering, base=None) -> float - [homogeneity_score](generated/cuml.metrics.cluster.homogeneity_score/index.html.md): cython_homogeneity_score(labels_true, labels_pred) -> float - [silhouette_score](generated/cuml.metrics.cluster.silhouette_score/index.html.md): Calculate the mean silhouette coefficient for the provided data. - [silhouette_samples](generated/cuml.metrics.cluster.silhouette_samples/index.html.md): Calculate the silhouette coefficient for each sample in the provided data. - [completeness_score](generated/cuml.metrics.cluster.completeness_score/index.html.md): cython_completeness_score(labels_true, labels_pred) -> float - [mutual_info_score](generated/cuml.metrics.cluster.mutual_info_score/index.html.md): cython_mutual_info_score(labels_true, labels_pred) -> float - [v_measure_score](generated/cuml.metrics.cluster.v_measure_score/index.html.md): cython_v_measure(labels_true, labels_pred, beta=1.0) -> float - [pairwise_distances](generated/cuml.metrics.pairwise_distances/index.html.md): Compute the distance matrix from a feature array X and optional Y. - [nan_euclidean_distances](generated/cuml.metrics.nan_euclidean_distances/index.html.md): Calculate the euclidean distances in the presence of missing values. - [pairwise_kernels](generated/cuml.metrics.pairwise_kernels/index.html.md): Compute the kernel between arrays X and optional array Y. - [cuml.model_selection](cuml.model_selection/index.html.md): This code is developed and maintained by scikit-learn and imported - [train_test_split](generated/cuml.model_selection.train_test_split/index.html.md): Split arrays or matrices into random train and test subsets. - [KFold](generated/cuml.model_selection.KFold/index.html.md): K-Folds cross-validator. - [cuml.multiclass](cuml.multiclass/index.html.md): | `OneVsOneClassifier` | Fit one binary classifier per pair of classes. | - [OneVsOneClassifier](generated/cuml.multiclass.OneVsOneClassifier/index.html.md): Fit one binary classifier per pair of classes. The input can be any kind - [OneVsRestClassifier](generated/cuml.multiclass.OneVsRestClassifier/index.html.md): Fit one binary classifier per class. The input can be any kind of cuML - [cuml.naive_bayes](cuml.naive_bayes/index.html.md): | `BernoulliNB` | Naive Bayes classifier for multivariate Bernoulli models. | - [BernoulliNB](generated/cuml.naive_bayes.BernoulliNB/index.html.md): Naive Bayes classifier for multivariate Bernoulli models. - [CategoricalNB](generated/cuml.naive_bayes.CategoricalNB/index.html.md): Naive Bayes classifier for categorical features. - [ComplementNB](generated/cuml.naive_bayes.ComplementNB/index.html.md): The Complement Naive Bayes classifier described in Rennie et al. (2003). - [GaussianNB](generated/cuml.naive_bayes.GaussianNB/index.html.md): Gaussian Naive Bayes (GaussianNB) - [MultinomialNB](generated/cuml.naive_bayes.MultinomialNB/index.html.md): Naive Bayes classifier for multinomial models. - [cuml.neighbors](cuml.neighbors/index.html.md): | `NearestNeighbors` | NearestNeighbors is an queries neighborhoods from a given set of data... - [NearestNeighbors](generated/cuml.neighbors.NearestNeighbors/index.html.md): NearestNeighbors is an queries neighborhoods from a given set of - [KNeighborsClassifier](generated/cuml.neighbors.KNeighborsClassifier/index.html.md): K-Nearest Neighbors Classifier is an instance-based learning technique, - [KNeighborsRegressor](generated/cuml.neighbors.KNeighborsRegressor/index.html.md): K-Nearest Neighbors Regressor is an instance-based learning technique, - [KernelDensity](generated/cuml.neighbors.KernelDensity/index.html.md): Kernel Density Estimation. Computes a non-parametric density estimate - [cuml.preprocessing](cuml.preprocessing/index.html.md): | `Binarizer` | Binarize data (set feature values to 0 or 1) according to a thre... - [Binarizer](generated/cuml.preprocessing.Binarizer/index.html.md): Binarize data (set feature values to 0 or 1) according to a threshold - [FunctionTransformer](generated/cuml.preprocessing.FunctionTransformer/index.html.md): Constructs a transformer from an arbitrary callable. - [KBinsDiscretizer](generated/cuml.preprocessing.KBinsDiscretizer/index.html.md): Bin continuous data into intervals. - [KernelCenterer](generated/cuml.preprocessing.KernelCenterer/index.html.md): Center a kernel matrix - [LabelBinarizer](generated/cuml.preprocessing.LabelBinarizer/index.html.md): Binarize labels in a one-vs-all fashion. - [LabelEncoder](generated/cuml.preprocessing.LabelEncoder/index.html.md): Encode target labels with values between 0 and n_classes - 1. - [MaxAbsScaler](generated/cuml.preprocessing.MaxAbsScaler/index.html.md): Scale each feature by its maximum absolute value. - [MinMaxScaler](generated/cuml.preprocessing.MinMaxScaler/index.html.md): Transform features by scaling each feature to a given range. - [MissingIndicator](generated/cuml.preprocessing.MissingIndicator/index.html.md): Binary indicators for missing values. - [Normalizer](generated/cuml.preprocessing.Normalizer/index.html.md): Normalize samples individually to unit norm. - [OneHotEncoder](generated/cuml.preprocessing.OneHotEncoder/index.html.md): Encode categorical features as a one-hot numeric array. - [OrdinalEncoder](generated/cuml.preprocessing.OrdinalEncoder/index.html.md): Encode categorical features as an integer array. - [PolynomialFeatures](generated/cuml.preprocessing.PolynomialFeatures/index.html.md): Generate polynomial and interaction features. - [PowerTransformer](generated/cuml.preprocessing.PowerTransformer/index.html.md): Apply a power transform featurewise to make data more Gaussian-like. - [QuantileTransformer](generated/cuml.preprocessing.QuantileTransformer/index.html.md): Transform features using quantiles information. - [RobustScaler](generated/cuml.preprocessing.RobustScaler/index.html.md): Scale features using statistics that are robust to outliers. - [SimpleImputer](generated/cuml.preprocessing.SimpleImputer/index.html.md): Imputation transformer for completing missing values. - [StandardScaler](generated/cuml.preprocessing.StandardScaler/index.html.md): Standardize features by removing the mean and scaling to unit variance - [TargetEncoder](generated/cuml.preprocessing.TargetEncoder/index.html.md): A cudf based implementation of target encoding [1], which converts - [add_dummy_feature](generated/cuml.preprocessing.add_dummy_feature/index.html.md): Augment dataset with an additional dummy feature. - [binarize](generated/cuml.preprocessing.binarize/index.html.md): Boolean thresholding of array-like or sparse matrix - [label_binarize](generated/cuml.preprocessing.label_binarize/index.html.md): Binarize labels in a one-vs-all fashion. - [maxabs_scale](generated/cuml.preprocessing.maxabs_scale/index.html.md): Scale each feature to the [-1, 1] range without breaking the sparsity. - [minmax_scale](generated/cuml.preprocessing.minmax_scale/index.html.md): Transform features by scaling each feature to a given range. - [normalize](generated/cuml.preprocessing.normalize/index.html.md): Scale input vectors individually to unit norm (vector length). - [robust_scale](generated/cuml.preprocessing.robust_scale/index.html.md): Standardize a dataset along any axis - [scale](generated/cuml.preprocessing.scale/index.html.md): Standardize a dataset along any axis - [cuml.random_projection](cuml.random_projection/index.html.md): | `GaussianRandomProjection` | Reduce dimensionality through Gaussian random projection. ... - [GaussianRandomProjection](generated/cuml.random_projection.GaussianRandomProjection/index.html.md): Reduce dimensionality through Gaussian random projection. - [SparseRandomProjection](generated/cuml.random_projection.SparseRandomProjection/index.html.md): Reduce dimensionality through sparse random projection. - [johnson_lindenstrauss_min_dim](generated/cuml.random_projection.johnson_lindenstrauss_min_dim/index.html.md): Find a ‘safe’ number of components to randomly project to. - [cuml.solvers](cuml.solvers/index.html.md): | `CD` | Coordinate Descent (CD) is a very common optimization algorithm that minimizes along coor... - [CD](generated/cuml.solvers.CD/index.html.md): Coordinate Descent (CD) is a very common optimization algorithm that - [QN](generated/cuml.solvers.QN/index.html.md): Quasi-Newton methods are used to either find zeroes or local maxima - [SGD](generated/cuml.solvers.SGD/index.html.md): Stochastic Gradient Descent is a very common machine learning algorithm - [cuml.svm](cuml.svm/index.html.md): | `SVC` | SVC (C-Support Vector Classification) | - [SVC](generated/cuml.svm.SVC/index.html.md): Construct an SVC classifier for training and predictions. - [SVR](generated/cuml.svm.SVR/index.html.md): * **Parameters:** - [LinearSVC](generated/cuml.svm.LinearSVC/index.html.md): Linear Support Vector Classification. - [LinearSVR](generated/cuml.svm.LinearSVR/index.html.md): Linear Support Vector Regression. ## Optional - [Top-level llms.txt](../llms.txt): Complete documentation index.