Resampling (statistics)

Resampling (English ) or sample repetition refers to the determination of the statistical properties of sample functions as estimators or test statistics based on a repeated drawing of samples, so-called sub-samples from a parent sample. The sampling function is repeatedly calculated on the basis of the proposed sub-samples and examined their distribution characteristics based on the results.

Computer support

Typically computer-based statistical analysis methods are used for resampling. They are needed because the probability distribution of a sample function or a statistical test can not always be determined ( at reasonable cost ). To specify in these situations confidence intervals and perform tests can, large numbers of ( pseudo-random ) data sets are based on the existing data with the help of simulation methods (Monte Carlo statistics) generated ( resampling ). These are then used to estimate the distribution of the sampling function, in particular their scattering parameters.

The methods have been developed since the 1980s. Known methods are Jackknife and Bootstrap.

Resampling method

Various methods are considered as resampling methods.

  • Bootstrapping ( statistics) parametric and non-parametric bootstrap
  • Bootstrap confidence intervals
  • Bootstrap tests

Applications

  • Bias correction and variance estimation
Sample rate conversion Pseudorandomness Jackknife (statistics) Bootstrapping (statistics)
678903
de