Calculating Shannon's entropy with JavaScript

entropy.js

/**
 * Calculate Shannon's entropy for a string
 */
module.exports = (str) => {
  const set = {};

  str.split('').forEach(
    c => (set[c] ? set[c]++ : (set[c] = 1))
  );

  return Object.keys(set).reduce((acc, c) => {
    const p = set[c] / str.length;
    return acc - (p * (Math.log(p) / Math.log(2)));
  }, 0);
};

Hmm, am not really that expert in Information Theory, but I wonder, can this implementation (seemingly simple), be trusted as legit? In my case, I wish to use this to compute the entropy of each string in a list, and pick the one with potentially the most information - am assuming that the higher the entropy, the more information a string has, is this correct?

@mcnemesis Yes.

Base-2 Shannon entropy calculation, it will give you a number from 0 to 8,
it is a simple frequency-calculation for each character, factorised into a specific range.

alternative that does not use the => syntax.

Nice. We can modernize and use ES6 set object:

const set = new Set();