Statistics & Binning

Statistics & Binning#

Containers that accumulate statistics incrementally, so that data can be pushed in several batches and partial results merged with +=. Use DescriptiveStatistics for moments over arrays, the binning and histogram classes to compute statistics on a grid, and TDigest for approximate quantiles.

Note

All the containers documented here can be populated in parallel from Dask arrays using the helpers in Dask Integration.

Descriptive Statistics#

DescriptiveStatistics(values[, weights, ...])

Univariate descriptive statistics.

Binning Containers#

Binning1D(x[, range, dtype])

Create a 1D binning for grouping values into bins on a vector.

Binning1DFloat32(self, x[, range])

Create a 1D binning for grouping values into bins on a vector.

Binning1DFloat64(self, x[, range])

Create a 1D binning for grouping values into bins on a vector.

Binning2D(x, y[, spheroid, dtype])

Create a 2D binning for grouping values into bins on a grid.

Binning2DFloat32(self, x, y[, spheroid])

Create a 2D binning for grouping values into bins on a grid.

Binning2DFloat64(self, x, y[, spheroid])

Create a 2D binning for grouping values into bins on a grid.

Histograms#

Histogram2D(x, y[, compression, dtype])

Create a 2D histogram for binning continuous values into a grid using TDigest.

Histogram2DFloat32(self, x, y[, compression])

Create a 2D histogram for binning continuous values into a grid using TDigest.

Histogram2DFloat64(self, x, y[, compression])

Create a 2D histogram for binning continuous values into a grid using TDigest.

Quantile Estimation#

TDigest(values[, weights, axis, ...])

T-Digest for incremental quantile estimation.

TDigestFloat32(self, values[, device, device])

T-Digest for incremental quantile estimation.

TDigestFloat64(self, values[, device, device])

T-Digest for incremental quantile estimation.