/**
 * Maximum-likelihood distribution fitting (Phase 4 Task 1).
 *
 * `fitDistribution(name, data)` fits one of five common distributions to a
 * sample by maximum likelihood and reports the fitted parameters plus the
 * achieved log-likelihood.
 *
 * - `normal`      — closed form: mu = mean, sigma = population std (ddof=0).
 * - `exponential` — closed form: lambda = 1 / mean.
 * - `lognormal`   — fit a normal to ln(data) (requires all data > 0).
 * - `poisson`     — closed form: lambda = mean.
 * - `gamma`       — no closed form for the shape parameter. Given shape k,
 *   the MLE scale is theta = xbar / k; substituting back yields the
 *   1-D shape equation `ln(k) - psi(k) = ln(xbar) - mean(ln x)`
 *   (psi = digamma), solved here with the secant method starting from the
 *   Choi & Wette (1969) initial guess.
 *
 * @packageDocumentation
 */
/** Supported distribution families for {@link fitDistribution}. */
export type DistributionName = 'normal' | 'exponential' | 'lognormal' | 'poisson' | 'gamma';
/** Result of {@link fitDistribution}: fitted parameters + achieved log-likelihood. */
export interface FitDistributionResult {
    /** Fitted parameters, named per distribution (see {@link fitDistribution}). */
    params: Record<string, number>;
    /** Log-likelihood of `data` under the fitted parameters. */
    logLikelihood: number;
}
/**
 * Fit a distribution to `data` by maximum likelihood.
 *
 * @param name - Distribution family: 'normal' | 'exponential' | 'lognormal' | 'poisson' | 'gamma'
 * @param data - Sample data
 * @returns Fitted parameters and the log-likelihood achieved under them
 *
 * @example
 * fitDistribution('normal', [2, 4, 4, 4, 5, 5, 7, 9]);
 * // { params: { mean: 5, std: 2 }, logLikelihood: ... }
 *
 * fitDistribution('exponential', [1, 2, 3, 2]);
 * // { params: { lambda: 0.5 }, logLikelihood: ... }
 */
export declare function fitDistribution(name: DistributionName, data: readonly number[]): FitDistributionResult;
//# sourceMappingURL=fit-distribution.d.ts.map