Determining the Statistical Power of the Kolmogorov-Smirnov and Anderson-Darling Goodness-of-Fit Tests via Monte Carlo Simulation

Abstract

Metrics are often used to compare the performance of newly developed systems with the performance of their predecessors. Metrics can also be used to compare the output of a simulator with real-world data to test the accuracy of the simulation.Statistical comparison of these metrics can be necessary when making such a determination. There are different methods of statistical comparison that are sensitive to the various types of underlying distribution of the metric data. Distribution type can affect the performance of these tests, and, fortunately, the distributions of many common metrics are well known. For example, mean time to repair (MTTR) and mean flight hours between critical failures (MFHBCF), generally follow a log-normal and an exponential distribution, respectively. This paper presents the effects of distribution type and parameters on the statistical power of two common goodness-of-fit tests (KolmogorovSmirnov and Anderson-Darling) via Monte Carlo simulation.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 2016
Accession Number
AD1029950

Entities

People

  • Brad M. Boyerinas

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Abstracts
  • Contracts
  • Data Science
  • Data Sets
  • Department Of Defense
  • Distribution Functions
  • Equations
  • Governments
  • Information Science
  • Instructions
  • Monte Carlo Method
  • National Governments
  • Normal Distribution
  • Simulations
  • Standards
  • Statistical Tests
  • United States Government

Fields of Study

  • Mathematics

Readers

  • Computational Modeling and Simulation
  • Inertial Navigation Systems.
  • Statistical inference.