import { RegressorBase } from "../base";
import { Params } from "../base/estimator";
import { SubsetSizeOption } from "../utils/paramResolvers";
interface RegressionTreeProps {
    max_depth?: number;
    min_samples_split?: number;
    max_features?: SubsetSizeOption;
    randomState?: number;
}
export declare class DecisionTreeRegressor extends RegressorBase {
    private feature_number;
    private regTree;
    private min_sample_split;
    private max_depth;
    private max_features?;
    private randomState?;
    private random;
    constructor(props?: RegressionTreeProps);
    getParams(): Params;
    private selectedFeatureIndices;
    /**
     * Minimizes total SSE = n_left * var_left + n_right * var_right over
     * midpoint thresholds, scanning each feature once in sorted order.
     */
    private attributeSelection;
    private initTreeNode;
    private buildTree;
    fit(sampleX: number[][], sampleY: number[]): void;
    private findLeaf;
    predict(sampleX: number[][]): number[];
    get featureImportances(): number[];
    getRawFeatureImportances(): number[];
    private collectLeaves;
    /**
     * Leaf index (in depth-first order) each sample falls into.
     */
    apply(sampleX: number[][]): number[];
    /**
     * Overwrite leaf predictions (ids as returned by apply). Gradient
     * boosting classifiers use this for the per-leaf Newton step.
     */
    setLeafValues(values: Map<number, number>): void;
}
export {};