import { ClassifierBase } from '../../base';
import { Params } from '../../base/estimator';
import { XGBoostProps } from './xgboostBase';
export interface XGBoostClassifierProps extends XGBoostProps {
}
/**
 * XGBoost classifier.
 *
 * Binary (K = 2): binary:logistic. p = sigmoid(F), g = p - y,
 * h = p * (1 - p); the initial margin is logit(base_score).
 *
 * Multiclass (K > 2): multi:softprob. Each round builds K trees on the
 * shared pre-round softmax probabilities with g = p_k - 1{y=k} and
 * h = max(2 * p_k * (1 - p_k), 1e-16), matching the xgboost library's
 * softmax objective (multiclass_obj: kEps = 1e-16f). Initial margins are 0 (softmax is shift-invariant,
 * so base_score has no effect here).
 */
export declare class XGBoostClassifier extends ClassifierBase {
    private params;
    private trees;
    private multiTrees;
    private baseMargin;
    private fitted;
    private classes;
    private nFeatures;
    constructor(props?: XGBoostClassifierProps);
    getParams(): Params;
    fit(trainX: number[][], trainY: number[]): void;
    private fitMulticlass;
    private predictMarginBinary;
    private predictMarginMulti;
    /**
     * Returns per-sample probabilities ordered by the sorted class labels.
     */
    predictProba(testX: number[][]): number[][];
    predict(testX: number[][]): number[];
    get featureImportances(): number[];
}