Standard

Transformer variant that maps named feature columns into a chosen value space. Each entry in transforms selects the space and log base for one column by name; columns that are not listed stay in linear space and are passed through unchanged.

Supported spaces:

• linear: identity (no transform).

• log: logarithm in the configured base. Inputs must be positive.

• asinh: inverse hyperbolic sine, scaled by the log of the base. Defined for all real inputs, so it tolerates zeros and negatives where log cannot.

Configuration

Selected as name: standard.

Option	Description	Type	Default
transforms	Mapping of feature-column name to a space selection.	dict[str, SpaceSelection]	{}

Each SpaceSelection has:

Option	Description	Type	Default
space	The value space to map the column into.	linear | log | asinh	required
base	Log base for log and asinh. Must be positive and not equal to 1.	int	10

components:
  transformer:
    name: standard
    kwargs:
      transforms:
        charge: { space: log, base: 10 }
        time:   { space: asinh, base: 10 }