KSLCore/ksl.utilities.distributions.fitting.scoring

Package-level declarations

Types

class AkaikeInfoCriterionScoringModel(val domain: Interval = DEFAULT_BIG_RANGE) : PDFScoringModel

Computes the Akaike Information Criterion (AIC) based on the data as the score. This assumes that the parameters of the supplied distribution have been estimated from the data and evaluates the likelihood associated with the current parameters of the distribution. The parameters of the distribution are not assumed to have been estimated from a maximum likelihood approach.

AndersonDarlingScoringModel

class AndersonDarlingScoringModel : PDFScoringModel

BayesianInfoCriterionScoringModel

class BayesianInfoCriterionScoringModel(val domain: Interval = DEFAULT_BIG_RANGE) : PDFScoringModel

Computes the Bayesian Information Criterion (BIC) based on the data as the score. This assumes that the parameters of the supplied distribution have been estimated from the data and evaluates the likelihood associated with the current parameters of the distribution. The parameters of the distribution are not assumed to have been estimated from a maximum likelihood approach.

ChiSquaredScoringModel

class ChiSquaredScoringModel : PDFScoringModel

CramerVonMisesScoringModel

class CramerVonMisesScoringModel : PDFScoringModel

KSScoringModel

class KSScoringModel : PDFScoringModel

MallowsL2ScoringModel

class MallowsL2ScoringModel : PDFScoringModel

This scoring model represents the Mallows L2 distance between the theoretical probabilities and the observed probabilities based on a histogram of the data. The break points for the histogram are specified by PDFModeler.equalizedCDFBreakPoints()

PDFScoringModel

abstract class PDFScoringModel(val name: String, val domain: Interval = Interval(0.0, Double.MAX_VALUE), val allowLowerLimitAdjustment: Boolean = true, val allowUpperLimitAdjustment: Boolean = true) : Metric

Computes a score to indicate the quality of fit for the proposed continuous distribution for the supplied data

PPCorrelationScoringModel

class PPCorrelationScoringModel(var empDistType: EmpDistType = EmpDistType.Continuity1, val domain: Interval = Interval(-1.0, 1.0)) : PDFScoringModel

This scoring model represents the linear correlation of a P-P plot for a fitted distribution. The correlation between the empirical probabilities and the theoretical probabilities for the data set is computed and returned as the score.

PPSSEScoringModel

class PPSSEScoringModel(var empDistType: EmpDistType = EmpDistType.Continuity1) : PDFScoringModel

This scoring model represents the sum of squared errors of a P-P plot for a fitted distribution. The sum of squared errors between the empirical probabilities and the theoretical probabilities for the data set is computed and returned as the score.

QQCorrelationScoringModel

class QQCorrelationScoringModel(var empDistType: EmpDistType = EmpDistType.Continuity1, val domain: Interval = Interval(-1.0, 1.0)) : PDFScoringModel

This scoring model represents the linear correlation of a Q-Q plot for a fitted distribution. The correlation between the empirical quantiles and the order statistics for the data set is computed and returned as the score.

QQSSEScoringModel

class QQSSEScoringModel(var empDistType: EmpDistType = EmpDistType.Continuity1) : PDFScoringModel

This scoring model represents the sum of squared errors of a Q-Q plot for a fitted distribution. The sum of squared difference between the empirical quantiles and the order statistics for the data set is computed and returned as the score.

SquaredErrorScoringModel

class SquaredErrorScoringModel : PDFScoringModel

This scoring model represents the sum of squared error between the predicted probabilities (based on the assumed distribution) and the observed probabilities. The break points for the histogram are specified by PDFModeler.equalizedCDFBreakPoints()