IterativeImputer

Multivariate imputation using chained equations (MICE algorithm) Compatible with sklearn.impute.IterativeImputer

Models each feature with missing values as a function of other features, and uses that estimate for imputation. It does so in an iterated round-robin fashion: at each step, a feature column is designated as output y and the other feature columns are treated as inputs X. A regressor is fit on (X, y) for known values and used to predict missing values of y.

Example

const imputer = new IterativeImputer({ max_iter: 10 });
imputer.fit(X_train);
const X_filled = imputer.transform(X_test);

Constructors

Constructor

new IterativeImputer(options?): IterativeImputer

Defined in: ds/src/ml/impute.js:761

Parameters

options?

copy

boolean = true

If true, create copy of X (default: true)

initial_strategy

string = "mean"

Initial imputation strategy (default: ‘mean’)

max_iter

number = 10

Maximum number of imputation rounds (default: 10)

max_value

number = Infinity

Maximum possible imputed value (default: Infinity)

min_value

number = -Infinity

Minimum possible imputed value (default: -Infinity)

tol

number = 1e-3

Tolerance for convergence (default: 1e-3)

verbose

boolean = false

Print progress (default: false)

Returns

IterativeImputer

Properties

_tableColumns

_tableColumns: any

Defined in: ds/src/ml/impute.js:779

copy

copy: boolean

Defined in: ds/src/ml/impute.js:776

initial_imputer_

initial_imputer_: SimpleImputer

Defined in: ds/src/ml/impute.js:778

initial_strategy

initial_strategy: string

Defined in: ds/src/ml/impute.js:770

max_iter

max_iter: number

Defined in: ds/src/ml/impute.js:771

max_value

max_value: number

Defined in: ds/src/ml/impute.js:774

min_value

min_value: number

Defined in: ds/src/ml/impute.js:773

n_iter_

n_iter_: number

Defined in: ds/src/ml/impute.js:780

nFeatures_

nFeatures_: number

Defined in: ds/src/ml/impute.js:777

tol

tol: number

Defined in: ds/src/ml/impute.js:772

verbose

verbose: boolean

Defined in: ds/src/ml/impute.js:775

Methods

_fitLinearRegression()

_fitLinearRegression(X, y): any

Defined in: ds/src/ml/impute.js:789

Fit a simple linear regression using pseudoinverse

Parameters

X

number[][]

Features

y

number[]

Target

Returns

any

Model with coefficients and predict function

_imputeFeature()

_imputeFeature(X, featureIdx, missing_mask): number[]

Defined in: ds/src/ml/impute.js:827

Impute a single feature using other features

Parameters

X

number[][]

Data matrix (current working copy, all values filled)

featureIdx

number

Index of feature to impute

missing_mask

boolean[][]

Original missingness mask

Returns

number[]

Imputed values for this feature

fit()

fit(X): IterativeImputer

Defined in: ds/src/ml/impute.js:907

Fit the imputer on training data

Parameters

X

any

Training data, table object, or {data, columns} format

Returns

IterativeImputer

this

fit_transform()

fit_transform(X): number[][]

Defined in: ds/src/ml/impute.js:1061

Fit and transform in one step

Parameters

X

any

Data to fit and transform

Returns

number[][]

Transformed data

transform()

transform(X): any[] | number[][]

Defined in: ds/src/ml/impute.js:941

Transform data by filling missing values using MICE

Parameters

X

any

Data to transform, table object, or {data, columns} format

Returns

any[] | number[][]

Transformed data (array if input was table)