import { BugRiskScore, TechDebtForecast, AnomalyEvent, MonteCarloResult, CodeMetrics, CodeAgeResult, ComplexityReport } from '../types.js';
/**
 * Predictive Engine — knows where bugs will appear before they do.
 *
 * Bug Risk Model: Weighted linear regression with exponential decay
 *   Features: churn rate, cyclomatic complexity, time since last test,
 *   number of authors, recent error frequency, code age
 *
 * Tech Debt Forecast: Time-series + seasonal decomposition
 *   Linear trend + seasonal component, 90-day projection
 *
 * Anomaly Detection: Z-score monitoring (rolling 7-day window, |z| > 2.5)
 *
 * Monte Carlo: 1000 simulations for debt payoff date estimation
 */
export declare class PredictiveEngine {
    private metricHistory;
    private readonly ANOMALY_WINDOW_DAYS;
    private readonly ANOMALY_Z_THRESHOLD;
    /**
     * Score bug risk for each file based on multiple factors.
     * Uses weighted linear regression with exponential time decay.
     */
    scoreBugRisk(files: CodeAgeResult[], metrics: CodeMetrics): BugRiskScore[];
    /**
     * Forecast tech debt trajectory using time-series analysis.
     * Fits linear trend + seasonal component, projects 90 days forward.
     */
    forecastDebt(history: ComplexityReport[]): TechDebtForecast;
    /**
     * Detect anomalies using Z-score on rolling window.
     * Alert when |z| > 2.5 (99.4% confidence interval).
     */
    detectAnomalies(metrics: number[], windowDays?: number): AnomalyEvent[];
    /**
     * Monte Carlo simulation for debt payoff estimation.
     * Runs N scenarios with randomized velocity to estimate completion dates.
     */
    monteCarlo(currentDebt: number, avgVelocity: number, simulations?: number): MonteCarloResult;
    /** Simple linear regression */
    private linearRegression;
    /** Box-Muller Gaussian random */
    private gaussianRandom;
}
//# sourceMappingURL=predictive-engine.d.ts.map