KSLCore/ksl.app.dist.result/GoodnessOfFitDTO

GoodnessOfFitDTO

@Serializable

data class GoodnessOfFitDTO(val chiSquaredStatistic: Double, val chiSquaredDOF: Int, val chiSquaredPValue: Double, val binBreakPoints: List<Double>, val binProbabilities: List<Double>, val expectedCounts: List<Double>, val observedCounts: List<Double>, val ksStatistic: Double? = null, val ksPValue: Double? = null, val andersonDarlingStatistic: Double? = null, val andersonDarlingPValue: Double? = null, val cramerVonMisesStatistic: Double? = null, val cramerVonMisesPValue: Double? = null, val indexOfDispersion: Double? = null, val poissonVarianceTestStatistic: Double? = null)(source)

Wire-safe goodness-of-fit results for one fitted distribution.

The chi-squared test and its bin table apply to both continuous and discrete fits. The continuous-only statistics (KS, Anderson-Darling, Cramer-von Mises) and the discrete-only statistics (index of dispersion, Poisson variance test) are nullable so a single DTO carries both kinds without forcing values that a given path does not compute.

The bin arrays (binBreakPoints, binProbabilities, expectedCounts, observedCounts) describe the chi-squared binning; they also give a client the exact bins the engine evaluated against.

Populated in a later phase by the result extractor; null until then.

Constructors

GoodnessOfFitDTO

constructor(chiSquaredStatistic: Double, chiSquaredDOF: Int, chiSquaredPValue: Double, binBreakPoints: List<Double>, binProbabilities: List<Double>, expectedCounts: List<Double>, observedCounts: List<Double>, ksStatistic: Double? = null, ksPValue: Double? = null, andersonDarlingStatistic: Double? = null, andersonDarlingPValue: Double? = null, cramerVonMisesStatistic: Double? = null, cramerVonMisesPValue: Double? = null, indexOfDispersion: Double? = null, poissonVarianceTestStatistic: Double? = null)