Schemassupervisedpreprocessor/v1

SupervisedPreprocessorConfig

Component

Language-independent config for the preprocessing a supervised run can fit: the preprocessing config without the operations a fitted preprocessor cannot replay at predict time (`complete_cases`, `remove_duplicates`, `remove_cases_thres`) or would learn differently in every resample (`remove_features_thres`). Those belong to `preprocessor`, applied to a dataset before training.

Raw JSONv1Unknown properties rejected

Properties

missingness

boolean

Add a boolean missingness indicator per feature with NAs.

impute

boolean

Impute missing values.

impute_type

string

Imputation method.

one of"missRanger""micePMM""meanMode"

impute_missRanger_params

object

Parameters passed to missRanger (e.g. pmm.k, maxiter, num.trees).

impute_discrete

string

Function name to impute discrete features.

impute_continuous

string

Function name to impute continuous features.

integer2factor

boolean

Convert integers to factors.

integer2numeric

boolean

Convert integers to numeric.

logical2factor

boolean

Convert logicals to factors.

logical2numeric

boolean

Convert logicals to numeric.

numeric2factor

boolean

Convert numeric to factors.

numeric2factor_levels

array | null≥ 1 items

Factor levels for numeric2factor.

numeric_cut_n

integer≥ 0

Cut numeric features into this many bins (0 = off).

numeric_cut_labels

boolean

Use labels for numeric_cut bins.

numeric_quant_n

integer≥ 0

Cut numeric features into this many quantile bins (0 = off).

numeric_quant_NAonly

boolean

Quantile-cut only features with NAs.

unique_len2factor

integer≥ 0

Convert features with <= this many unique values to factors (0 = off).

character2factor

boolean

Convert character features to factors.

factorNA2missing

boolean

Convert factor NAs to a 'missing' level.

factorNA2missing_level

string

Level name for factorNA2missing.

factor2integer

boolean

Convert factors to integers.

factor2integer_startat0

boolean

factor2integer starts at 0.

factor2integer_levels

object | null

Per-feature factor2integer levels, keyed by feature name.

scale

boolean

Scale features.

center

boolean

Center features.

scale_centers

object | null

Per-feature centering values, keyed by feature name.

scale_coefficients

object | null

Per-feature scaling values, keyed by feature name.

remove_constants

boolean

Remove constant features.

remove_constants_skip_missing

boolean

Ignore missing values when detecting constants.

remove_features

array | null≥ 1 items

Names of features to remove.

one_hot

boolean

One-hot encode factors.

one_hot_levels

object | null

Per-feature one-hot levels, keyed by feature name.

add_date_features

boolean

Add date-derived features.

date_features

string[]≥ 1 items

Date features to add.

add_holidays

boolean

Add a holiday indicator feature.

holidays

string[]≥ 1 itemsunique items

Holidays to flag, named as timeDate holiday functions.

exclude

array | null≥ 1 items

Column indices to exclude from preprocessing.

Relationships

Used by

supervised