bootstrap

bootstrap

In statistics the Bootstrap is a method to estimate the statistical error of observables measured on a set of data.

References

• http://en.wikipedia.org/wiki/Bootstrapping

bootstrap(cfgs,n,f)

This function implements the Bootstrap method for functions of a set of vector data saved as the rows of a matrix. The function has to take a matrix of data as the first argument and has to return either a scalar or a matrix.

Argument cfgs

Matrix

Argument n

Scalar

Argument f

Function

Returns 1. entry

Matrix

Example

Bootstrap([3 + randn(100, 1), 10 + 2 * randn(100, 1)], 200, avg)

Gives the statistical Bootstrap estimate for the mean and the error on the mean of a dataset with 100 measurements, mean [3, 10], and gaussian noise of width [1, 4] for 200 bootstrap samples.

bootstrap(cfgs,n,f,P)

This function implements the Bootstrap method for functions of a set of vector data saved as the rows of a matrix. The function has to take a matrix of data as the first argument, can take constant other optional arguments, and has to return either a scalar or a matrix.

Argument cfgs

Matrix

Argument n

Scalar

Argument f

Function

Argument P

Arguments

Returns 1. entry

Matrix

Example

Bootstrap([3 + randn(100, 1), 10 + 2 * randn(100, 1)], 200, avg)

Gives the statistical Bootstrap estimate for the mean and the error on the mean of a dataset with 100 measurements, mean [3, 10], and gaussian noise of width [1, 4] for 200 bootstrap samples.