Covariance and Correlation in Javascript

stats.js

function sum(xs) {
  return xs.reduce((total, x) => total + x, 0)
}

function mean(xs) {
  return sum(xs) / xs.length
}

function zip(xs, ys) {
  return range(xs.length)
    .reduce((arr, i) => [...arr, [xs[i], ys[i]]], [])
}

function range(start, end) {
  if (end == null) {
    end = start
    start = 0
  }
  
  return new Array(end - start).fill().map((_, i) => i + start)
}

function stdev(xs) {
  const xhat = mean(xs)
  const squareDistances = xs.map(x => Math.pow(x - xhat, 2))
  return Math.sqrt(sum(squareDistances) / (xs.length - 1))
}

function covariance(xs, ys) {
  const xhat = mean(xs)
  const yhat = mean(ys)
  
  const total = sum(zip(xs, ys).map(([x, y]) => (x - xhat) * (y - yhat)))
  
  return total / (xs.length - 1)
}

function correlation(xs, ys) {
  return covariance(xs, ys) / (stdev(xs) * stdev(ys))
}

const xs = [1, 2, 3, 4, 5]
console.log(stdev(xs)) // 1.5811388300841898

const ys = [1, 2, 3, 4, 5]
console.log(covariance(xs, ys)) // 2.5
console.log(correlation(xs, ys)) // 0.9999999999999998

function rand(size) {
  return range(size).map(() => Math.random() * size)
}

console.log(correlation(rand(100), rand(100))) // (very small +ve/-ve number)