Resampling
Resampling or sample repetition refers to the determination of the statistical properties of sample functions , such as estimators or test variables, on the basis of a repeated drawing of samples, so-called sub-samples, from an initial sample. The sample function is calculated repeatedly on the basis of the sub-samples drawn and its distribution properties are examined on the basis of the results.
Computer support
Computer-aided statistical evaluation methods are typically used for resampling. They are needed because the probability distribution of a sample function or a statistical test cannot always be determined (with reasonable effort). In order to be able to specify confidence intervals and carry out tests in these situations as well, large numbers of ( pseudo-random ) data records are generated (resampling) on the basis of the available data with the aid of simulation methods ( Monte Carlo statistics ). These are then used to estimate the distribution of the sample function, especially its dispersion parameters.
The procedures have been developed since the 1980s. Known methods are the jackknife method and the bootstrapping method .
Resampling process
Various methods are counted among the resampling methods.
-
Bootstrapping process
- parametric and non-parametric bootstrap
- Bootstrap confidence ranges
- Bootstrap tests
- Jackknife method
- Cross validation
- Permutation tests or randomization tests
- RANSAC algorithm
Applications
Individual evidence
- ↑ Bernd Rönz, Hans G. Strohe (1994), Lexicon Statistics , Gabler Verlag, p. 312.
literature
- Y. Shao, D. Tu: The Jackknife and Bootstrap. Springer, New York, 1995
- B. Efron, RG Tibshirani: An Introduction to the Bootstrap. Chapman and Hall, New York, 1993
- EF Harrell: Regression Modeling Strategies With Applications to Linear Models, Logistic Regression, and Survival Analysis, Springer, New York, 2006