histogram() function

histogram() approximates the cumulative distribution of a dataset by counting data frequencies for a list of bins.

A bin is defined by an upper bound where all data points that are less than or equal to the bound are counted in the bin. Bin counts are cumulative.

Each input table is converted into a single output table representing a single histogram. Each output table has the same group key as the corresponding input table. Columns not part of the group key are dropped. Output tables include additional columns for the upper bound and count of bins.

Function type signature

(
    <-tables: stream[A],
    bins: [float],
    ?column: string,
    ?countColumn: string,
    ?normalize: bool,
    ?upperBoundColumn: string,
) => stream[B] where A: Record, B: Record

Parameters

column

Column containing input values. Column must be of type float. Default is _value.

upperBoundColumn

Column to store bin upper bounds in. Default is le.

countColumn

Column to store bin counts in. Default is _value.

bins

(Required) List of upper bounds to use when computing the histogram frequencies.

Bins should contain a bin whose bound is the maximum value of the data set. This value can be set to positive infinity if no maximum is known.

Bin helper functions

The following helper functions can be used to generated bins.

• linearBins()
• logarithmicBins()

normalize

Convert counts into frequency values between 0 and 1. Default is false.

Note: Normalized histograms cannot be aggregated by summing their counts.

tables

Input data. Default is piped-forward data (<-).

Examples

• Create a cumulative histogram
• Create a cumulative histogram with dynamically generated bins

Create a cumulative histogram

import "sampledata"

sampledata.float()
    |> histogram(bins: [0.0, 5.0, 10.0, 20.0])

Create a cumulative histogram with dynamically generated bins

import "sampledata"

sampledata.float()
    |> histogram(bins: linearBins(start: 0.0, width: 4.0, count: 3))