Bootstrapping process

from Wikipedia, the free encyclopedia

The bootstrapping method or bootstrap method (rarely Münchhausen method ) is a method of resampling in statistics . Statistics are repeatedly calculated based on just a sample. Bootstrap methods are used when the theoretical distribution of the statistics of interest is not known. This method was first described by Bradley Efron in 1979 in Bootstrap Methods: Another Look at the Jackknife .

The bootstrap usually replaces the theoretical distribution function of a random variable with the empirical distribution function (relative cumulative frequency function) of the sample .

Action

For this, in the simplest case, bootstrap samples are generated by taking a value with replacement from the given sample for each drawing . This corresponds to the repeated extraction of random numbers from the empirical distribution function . For each bootstrap sample, the value of the statistic of interest is calculated. The distribution of is finally approximated by the empirical distribution of the values .

In less intuitive models, it is not just repeated pulling from the data that is already available. In terms of method, the bootstrap method can also proceed in such a way that certain parameters of the unknown distribution are estimated and data is regenerated on the basis of this information by generating a distribution with the estimated values.

Web links

literature

  • Bradley Efron: Bootstrap Methods: Another Look at the Jackknife . In: The Annals of Statistics . 7, No. 1, 1979, pp. 1-26. doi : 10.1214 / aos / 1176344552 .
  • B. Efron, RJ Tibshirani: An introduction to the bootstrap , New York: Chapman & Hall, 1993
  • J. Shao, D. Tu: The Jackknife and Bootstrap . Springer, 1995.