Skip to main content
torch.js has not been released yet.
torch.js logotorch.js logotorch.js
PlaygroundContact
Login
Documentation
IntroductionType SafetyTensor ExpressionsTensor IndexingEinsumEinopsAutogradTraining a ModelProfiling & MemoryPyTorch MigrationBest PracticesRuntimesPerformancePyTorch CompatibilityBenchmarksDType Coverage
IntroductionRenderer GalleryRenderersAnalysis
computeQuantilescomputePercentilescomputeCorrelationcomputeCovariancedetectOutliersdetectOutliersZScorecomputeExtendedStatsdescribeExtendedStatsQuantileResultCorrelationResultOutlierResultExtendedStatscreateDBSCANdbscanDBSCANDBSCAN.fitDBSCAN.destroyDBSCANOptionsDBSCANResultcreateHierarchicalClusteringhierarchicalClusteringHierarchicalClusteringHierarchicalClustering.fitHierarchicalClustering.destroyLinkageMethodDistanceMetricHierarchicalOptionsDendrogramNodeHierarchicalResultcomputeHistogramgetHistogramCentersnormalizeHistogramHistogramResultcreateKDEkdeKDEKDE.fitKDE.destroyKernelTypeKDEOptionsKDEResultcreateKMeanskmeansKMeansKMeans.fitKMeans.destroyKMeansOptionsKMeansResultcreatePCApcaPCAPCA.fitPCA.destroyPCAOptionsPCAResultthrottleProgressconsoleProgressProgressInfoProgressCallbackcomputeStatscomputeMinMaxdescribeStatsTensorStatscreateTSNEtsneTSNETSNE.fitTSNE.destroyTSNEOptionsTSNEResultcreateUMAPumapUMAPUMAP.fitUMAP.destroyUMAPOptionsUMAPResult
torch.js· 2026
LegalTerms of UsePrivacy Policy
/
/
  1. docs
  3. viz
  5. viz
  7. analysis

viz.analysis

Functions

  • computeQuantiles - Compute quantiles (percentiles) of a tensor.
  • computePercentiles - Compute common percentiles (5th, 25th, 50th, 75th, 95th).
  • computeCorrelation - Compute Pearson correlation matrix for multiple variables.
  • computeCovariance - Compute covariance matrix for multiple variables.
  • detectOutliers - Detect outliers using the Interquartile Range (IQR) method.
  • detectOutliersZScore - Detect outliers using Z-score method.
  • computeExtendedStats - Compute extended statistics including skewness and kurtosis.
  • describeExtendedStats - Describe extended statistics as a formatted string.
  • createDBSCAN - Create a new DBSCAN instance.
  • dbscan - Convenience function for one-shot DBSCAN clustering.
  • createHierarchicalClustering - Create a new HierarchicalClustering instance.
  • hierarchicalClustering - Convenience function for one-shot hierarchical clustering.
  • computeHistogram - Compute a histogram of tensor values on GPU.
  • getHistogramCenters - Generate bin centers from histogram result.
  • normalizeHistogram - Normalize histogram counts to sum to 1 (probability density).
  • createKDE - Create a new KDE instance.
  • kde - Convenience function for one-shot KDE computation.
  • createKMeans - Create a new KMeans clustering instance.
  • kmeans - One-shot k-means clustering with automatic resource cleanup.
  • createPCA - Create a new PCA instance.
  • pca - Convenience function for quick PCA projection.
  • throttleProgress - Create a throttled progress callback that only fires at most every ms milliseconds.
  • consoleProgress - Create a progress callback that logs to console.
  • computeStats - Compute all statistics (min, max, mean, std, sum) entirely on GPU.
  • computeMinMax - Compute just min and max values (for performance when other stats not needed).
  • describeStats - Generate human-readable description of tensor statistics.
  • createTSNE - Create a new TSNE instance.
  • tsne - Convenience function to run t-SNE.
  • createUMAP - Create a new UMAP instance.
  • umap - Convenience function to run UMAP.

DBSCAN

  • fit - Fit DBSCAN clustering on data.
  • destroy - Clean up GPU resources.

HierarchicalClustering

  • fit - Fit hierarchical clustering on data.
  • destroy - Clean up GPU resources.

KDE

  • fit - Compute KDE (Kernel Density Estimation) for 1D data.
  • destroy - Clean up GPU resources.

KMeans

  • fit - Fit k-means clustering to data.
  • destroy - Clean up GPU resources.

PCA

  • fit - Fit PCA and transform data to lower dimensions.
  • destroy - Clean up GPU resources.

TSNE

  • fit - Fit t-SNE to data.
  • destroy - Clean up resources.

UMAP

  • fit - Fit UMAP to input data.
  • destroy - Clean up resources.

Types

  • QuantileResult - Result from quantile computation.
  • CorrelationResult - Result from correlation computation.
  • OutlierResult - Result from outlier detection.
  • ExtendedStats - Extended statistics including higher-order moments.
  • DBSCANOptions - Options for DBSCAN clustering.
  • DBSCANResult - Result of DBSCAN clustering.
  • LinkageMethod - Linkage method for hierarchical clustering.
  • DistanceMetric - Distance metric for hierarchical clustering.
  • HierarchicalOptions - Options for hierarchical clustering.
  • DendrogramNode - A single merge in the dendrogram.
  • HierarchicalResult - Result of hierarchical clustering.
  • HistogramResult -
  • KernelType - Kernel function types for KDE.
  • KDEOptions - Options for KDE computation.
  • KDEResult - Result of KDE computation.
  • KMeansOptions -
  • KMeansResult -
  • PCAOptions -
  • PCAResult -
  • ProgressInfo - Progress callback types for long-running analysis operations.
  • ProgressCallback - Progress callback function type.
  • TensorStats -
  • TSNEOptions -
  • TSNEResult -
  • UMAPOptions - Options for UMAP computation.
  • UMAPResult - Result from UMAP computation.