jackknife_error: Estimates error from jackknife samples
In hadron: Analysis Framework for Monte Carlo Simulation Data in Physics
jackknife_error R Documentation
Estimates error from jackknife samples
Description
Computes the jackknife error which is just
∑_{i=0}^N (x_i - \bar x)^2 \,.
Internally we use
\frac{(N-1)^2}{N} \mathop{\mathrm{sd}}(X)
in order to benefit from the optimized standard deviation function.
The width of the bootstrap distribution does not change with the number of
elements. The jackknife distribution crucially depends on the number of
measurements that one started with. Therefore we cannot just drop the NA
values and are done with it. Instead we need to rescale with the
√{N / m} where N is the number of original measurements and
m is the number of non-NA values. With NA values removed we would
otherwise underestimate the uncertainty.
Usage
jackknife_error(samples, boot.l = 1, na.rm = FALSE)
Arguments
samples
Numeric vector.
boot.l
Block length for bootstrapping.
na.rm
Logical. Determines whether NA values shall be removed, see
Description for details.
Details
Currently this uses the mean over the jackknife samples in order to compute
the error. It would be better in the case of a bias to use the mean over the
original data instead. This would require a second parameter and therefore
is incompatible with the previously used sd everywhere for the bootstrap
samples. As the sd for the bootstrap samples also does not include the
original data, this likely is similar in terms of bias.
Value
returns a single numeric representing the jackknife estimate of error
hadron documentation built on Sept. 9, 2022, 5:06 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
| jackknife_error | R Documentation |
Estimates error from jackknife samples
Description
Computes the jackknife error which is just
∑_{i=0}^N (x_i - \bar x)^2 \,.
Internally we use
\frac{(N-1)^2}{N} \mathop{\mathrm{sd}}(X)
in order to benefit from the optimized standard deviation function.
The width of the bootstrap distribution does not change with the number of elements. The jackknife distribution crucially depends on the number of measurements that one started with. Therefore we cannot just drop the NA values and are done with it. Instead we need to rescale with the √{N / m} where N is the number of original measurements and m is the number of non-NA values. With NA values removed we would otherwise underestimate the uncertainty.
Usage
jackknife_error(samples, boot.l = 1, na.rm = FALSE)
Arguments
samples |
Numeric vector. |
boot.l |
Block length for bootstrapping. |
na.rm |
Logical. Determines whether |
Details
Currently this uses the mean over the jackknife samples in order to compute
the error. It would be better in the case of a bias to use the mean over the
original data instead. This would require a second parameter and therefore
is incompatible with the previously used sd everywhere for the bootstrap
samples. As the sd for the bootstrap samples also does not include the
original data, this likely is similar in terms of bias.
Value
returns a single numeric representing the jackknife estimate of error
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.