import { BaseEstimator, Params } from '../base/estimator';
import { SingleOutputEstimator } from './common';
export interface MultiOutputClassifierProps {
    /** Prototype classifier; one unfitted clone is fitted per output column. */
    estimator: BaseEstimator;
}
/**
 * Multi-target classification wrapper, mirroring sklearn's
 * `MultiOutputClassifier`: one clone of `estimator` is fitted per output
 * column of the 2-D target matrix `Y` [nSamples][nOutputs].
 *
 * `fit(X, Y)` takes a 2-D target, so this extends `BaseEstimator` directly
 * rather than `ClassifierBase` (whose contract is 1-D y).
 *
 * `score(X, Y)` is **subset accuracy** (exact-match): the fraction of samples
 * whose *entire* output row is predicted correctly — sklearn's
 * `MultiOutputClassifier.score`, which applies `accuracy_score` to 2-D
 * targets. This is deliberately harsher than the mean of per-output
 * accuracies.
 *
 * Nested params are addressable as `estimator__<param>` in `setParams`.
 */
export declare class MultiOutputClassifier extends BaseEstimator {
    private estimator;
    private estimators_;
    constructor(props: MultiOutputClassifierProps);
    getParams(): Params;
    /** Supports own params plus nested `estimator__param` addressing. */
    setParams(params: Params): this;
    /** The fitted per-output members, ordered like Y's columns. */
    get estimators(): SingleOutputEstimator[];
    fit(X: number[][], Y: number[][], sampleWeight?: number[]): void;
    private ensureFitted;
    /** Predicted target matrix, shape [nSamples][nOutputs]. */
    predict(X: number[][]): number[][];
    /**
     * Per-output class-membership probabilities, shape
     * [nOutputs][nSamples][nClasses(output)] — one probability matrix per
     * output, like sklearn's list of arrays. Requires the members to
     * implement `predictProba`.
     */
    predictProba(X: number[][]): number[][][];
    /**
     * Subset (exact-match) accuracy: the fraction of rows where every output
     * is predicted correctly. Matches sklearn's `MultiOutputClassifier.score`.
     */
    score(X: number[][], Y: number[][]): number;
}