ExponentialScale

class

class ExponentialScale(
    min_val: ScaleLimitSpec,
    max_val: ScaleLimitSpec,
    base: ScaleBaseSpec = math.e,
    bins: BinCountSpec = MAX_BINS,
    use_bin_zero: bool = False,
)

Bases: Scale

Categories: dq-classifier

A scale whose bin edges are spaced exponentially over [min_val, max_val] (used with ScaleCategorizer).

The edges grow geometrically, so the bins widen as values grow, the mirror of LogarithmicScale. A range with min_val <= 0 is shifted to positive values before the edges are computed.

Examples

Bins that widen as values grow -- useful for long-tailed values. Here the edges are 10, 100 and 1000:

ScaleCategorizer(
    ["amount"],
    scale=ExponentialScale(1.0, 1000.0, base=10.0, bins=3),
)

Parameters

parameter

min_valScaleLimitSpec (StrictInt | StrictFloat)

Lowest value of the scale; an int or float.

parameter

max_valScaleLimitSpec (StrictInt | StrictFloat)

Highest value of the scale; strictly greater than min_val.

parameter

baseScaleBaseSpec (StrictFloat)

Base of the exponent; a float greater than 0 (default e).

parameter

binsBinCountSpec (StrictInt)

Number of bins; a whole number from 1 to 100 (default 100).

parameter

use_bin_zerobool

If False, the minimum value belongs to the first bin; if True, min_val gets its own bin 0.