Scribd
Upload
3 views8 pages

Resampling Techniques in Statistics

The document discusses resampling techniques, specifically bootstrapping and jackknifing, which are used when the distributions of estimated parameters are unknown. Bootstrapping involves random resampling with replacement to estimate bias, standard error, and confidence intervals, while jackknifing does not involve random sampling and focuses on recalculating statistics by holding out one observation at a time. The document highlights the assumptions, equations, potential problems, and provides an example of applying these techniques in multiple linear regression.

Uploaded by

Neeraj Yadav
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
3 views8 pages

Resampling Techniques in Statistics

The document discusses resampling techniques, specifically bootstrapping and jackknifing, which are used when the distributions of estimated parameters are unknown. Bootstrapping involves random resampling with replacement to estimate bias, standard error, and confidence intervals, while jackknifing does not involve random sampling and focuses on recalculating statistics by holding out one observation at a time. The document highlights the assumptions, equations, potential problems, and provides an example of applying these techniques in multiple linear regression.

Uploaded by

Neeraj Yadav
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd

Resampling

Bootstrapping
Jackknifing

03/21/25 AGR206 1
Purpose of resampling
 When the distributions of the
estimated parameters are not known,
either because
 The distribution of the original
variable is not known, or
 There is no mathematical formula or
rationale to relate the distribution of
the estimator to the distribution of the
original variables.
 Example:
 there is no “formula” for the
distribution of the product of two
normal distributions!

2
03/21/25 AGR206
When to use resampling
 When distributions are unknown/not met.
 Robust regression.
 Weighted regression.

 When formulas are unknown/cumbersome.


 Unusual ecological indices.

3
03/21/25 AGR206
Common resampling techniques
 BOOTSTRAPPING  JACKKNIFING
 Random resampling  No random
with replacement. component.
 Gives bias, s. error  Gives bias & s. error,
and CI’s. but no CI.
 Uses sample as proxy  Hold out one
of population and observation at a time
repeats sampling and recalculate
process exactly the statistics for each
same way as for the subset.
original sample to
obtain hundreds of
bootstrap samples.

4
03/21/25 AGR206
Assumptions:
 Observations of the original sample
are independent.
 Sample is not small (n>30).

5
03/21/25 AGR206
Bootstrapping equations:
 Take a statistic π(x) where x are the
values of a measure in the sample. For
example when π(x)=x/n, π is the
average.
 Let x denote the original sample and xi*
denote a bootstrap sample.
 Average bias=average π(xi*)-π(x).
 Estimated variance
 S2π= [π(xi*)-avg π(xi*)]2/(B-1)
where B is the number of bootstrap
samples.

6
03/21/25 AGR206
Potential problems
 Small sample size.
 Computationally demanding.
 Lack of user-friendly software for
general purpose.
 Bootstrapping for CI requires many
more bootstrap samples and has to
be corrected for bias.

7
03/21/25 AGR206
Example

 Multiple Linear regression


 Fixed X case. Resample from the
residuals and add to predicted
values.
 Random X case. Resample from the
observations, taking each as a unit.

8
03/21/25 AGR206

You might also like