resample

Resampling methods for variance and bias estimation. Currently implemented such techniques:

• Resample
• Shuffle
• Bootstrap
• Jackknife

no tests yet

resample generates new samples from an input array and applies to them passed function to generate multiple estimation replicates. Then those replicates are used to calculate mean, bias and standard error of an estimate

Install

npm i -S resample

Usage

var { bootstrap, jackknife } = require('resample')
 
function mean (array) {
  return array.reduce((a, b) => a + b) / array.length
}
 
bootstrap([1, 2, 3, 4, 5, 6, 40], mean)
// { observed: 8.71, mean: 8.74, se: 4.84, bias: 0.029, replicates: [...], samples: [...] }
 

The bootstrap method supports an extra parameter that sets number of samples/iterations (default: 10000)

bootstrap([1, 2, 3, 4, 5, 6, 40], mean, 10000)

Passing only number of samples without the estimator function return raw samples with replacement

bootstrap([1, 2, 3, 4, 5, 6, 40], 2))
// [[ 2, 40, 1, 1, 5, 2, 6], [2, 3, 4, 4, 2, 40, 6]]