Function Reference: jackknife

statistics: jackstat = jackknife (E, x)
statistics: jackstat = jackknife (E, x, …)

Compute jackknife estimates of a parameter taking one or more given samples as parameters.

In particular, E is the estimator to be jackknifed as a function name, handle, or inline function, and x is the sample for which the estimate is to be taken. The i-th entry of jackstat will contain the value of the estimator on the sample x with its i-th row omitted.

 
 
 jackstat (i) = E(x(1 : i - 1, i + 1 : length(x)))

Depending on the number of samples to be used, the estimator must have the appropriate form:

  • If only one sample is used, then the estimator need not be concerned with cell arrays, for example jackknifing the standard deviation of a sample can be performed with jackstat = jackknife (@std, rand (100, 1)).
  • If, however, more than one sample is to be used, the samples must all be of equal size, and the estimator must address them as elements of a cell-array, in which they are aggregated in their order of appearance:
 
 
 jackstat = jackknife (@(x) std(x{1})/var(x{2}),
 rand (100, 1), randn (100, 1))

If all goes well, a theoretical value P for the parameter is already known, n is the sample size,

t = n * E(x) - (n - 1) * mean(jackstat)

and

v = sumsq(n * E(x) - (n - 1) * jackstat - t) / (n * (n - 1))

then

(t-P)/sqrt(v) should follow a t-distribution with n-1 degrees of freedom.

Jackknifing is a well known method to reduce bias. Further details can be found in:

References

  1. Rupert G. Miller. The jackknife - a review. Biometrika (1974), 61(1):1-15. doi:10.1093/biomet/61.1.1
  2. Rupert G. Miller. Jackknifing Variances. Ann. Math. Statist. (1968), Volume 39, Number 2, 567-582. doi:10.1214/aoms/1177698418

Source Code: jackknife

Example: 1

  

 for k = 1:1000
   rand ("seed", k);  # for reproducibility
   x = rand (10, 1);
   s(k) = std (x);
   jackstat = jackknife (@std, x);
   j(k) = 10 * std (x) - 9 * mean (jackstat);
 endfor
 figure();
 hist ([s', j'], 0:sqrt(1/12)/10:2*sqrt(1/12))
plotted figure

Example: 2

  

 for k = 1:1000
   randn ("seed", k); # for reproducibility
   x = randn (1, 50);
   rand ("seed", k);  # for reproducibility
   y = rand (1, 50);
   jackstat = jackknife (@(x) std(x{1})/std(x{2}), y, x);
   j(k) = 50 * std (y) / std (x) - 49 * mean (jackstat);
   v(k) = sumsq ((50 * std (y) / std (x) - 49 * jackstat) - j(k)) / (50 * 49);
 endfor
 t = (j - sqrt (1 / 12)) ./ sqrt (v);
 figure();
 plot (sort (tcdf (t, 49)), ...
       "-;Almost linear mapping indicates good fit with t-distribution.;")
plotted figure