Development build. This page describes main, not a released package. The latest published Lodestar.Survival is 0.2.0 — read its documentation.

Home › Survival › Survival estimators

AalenOptions

What an AalenAdditive fit may be told.

public sealed record AalenOptions

Properties — ConfidenceLevel is the two-sided level the cumulative coefficients' bounds are reported at; 0.95 by default. FitIntercept is lifelines' fit_intercept: whether a baseline column of ones is added after the covariates, true by default. CoefficientPenalizer is lifelines' coef_penalizer, a ridge penalty on each time's increments, zero by default. SmoothingPenalizer is lifelines' smoothing_penalizer, which pulls each time's increments towards the previous time's, zero by default.

Exceptions — ArgumentOutOfRangeException when ConfidenceLevel does not lie strictly inside (0, 1), or when either penalizer is negative, infinite or NaN. Each is thrown where the setting is set, not where the fit reads it.

Example — a ridge penalty shrinks the treatment's cumulative coefficient at the last event time.

using Lodestar.Survival;

double[] treated = [0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1];
double[] months = [3, 10, 5, 4, 8, 18, 12, 6, 9, 20, 7, 15];
bool[] died = [true, true, true, false, true, true, true, true, true, false, true, true];

AalenSummary plain = AalenAdditive.Fit(treated, months, died, 1);
AalenSummary ridge = AalenAdditive.Fit(
    treated, months, died, 1, new AalenOptions { CoefficientPenalizer = 1.0 });

int last = (plain.EventTimes.Count - 1) * 2;
double unpenalised = Math.Round(plain.CumulativeHazards[last], 6);   // => -2.083333
double penalised = Math.Round(ridge.CumulativeHazards[last], 6);     // => -0.643275

Remarks — both penalties act on the scaled covariates, each at unit sample deviation, so they do not depend on the units a covariate is recorded in. At each event time the two add to the diagonal of the normal equations together, and the smoothing penalty adds its share of the previous time's increments to the right-hand side: at CoefficientPenalizer = 0, a large SmoothingPenalizer holds the increments near their previous values, a small one leaves each time to its own deaths. The intercept is penalised with the rest.

Applies to — net10.0, netstandard2.0.

See also — AalenAdditive.Fit, AalenSummary.