Basic descriptive statistics using Ruby

stats.rb

# The _percentile_sorted function is adapted from similar code
# released under the Academic Free License, 3.0, written by
# Derrick Pallas, which used to be available at http://derrick.pallas.us/ruby-stats/
# it fixes a bug that can occur if the list of numbers is small.
# Other code more or less from my [first ever gist](https://gist.github.com/willf/187846)

def _percentile_sorted(p, ns)
  r = p / 100.00 * (ns.size + 1)
  i, fr = r.divmod(1)
  (fr==0.0 || i >= ns.size) ? ns[i-1] : ns[i-1] + (fr * (ns[i]-ns[i-1]))
end

# This requires a O(n log n) sort of the items in the array
# Plus a one time pass through the array
# and a one time pass through the distinct values of the array
def stats(ns, include_histogram=true)
  ns = ns.sort
  len = ns.size
  histogram = Hash.new(0)
  n, mean, m2, sum = ns.inject([0,0.0,0.0,0.0]) do
    |acc, r|
    n = acc[0]+1
    delta = r-acc[1]
    mean = acc[1] + (delta/n)
    m2 = acc[2] + (delta * (r-mean))
    sum = acc[3] + r
    histogram[r]+=1
    [n, mean, m2, sum]
  end
  var = m2/n
  max_count = histogram.values.max
  {
    n: len,
    mean: mean,
    variance: var,
    stddev: Math.sqrt(var),
    sum: sum,
    min: ns.first,
    max: ns.last,
    histogram: (include_histogram ? histogram : {}),
    modes: include_histogram ? histogram.select{|x| histogram[x]==max_count}: {},
    median: (len % 2 == 1) ? ns[len/2] : (ns[len/2]+(ns[len/2-1])/2),
    percentiles: [10, 20, 30, 40, 50, 60, 70, 80, 90, 95, 99, 99.9].inject({}){|h,x| h[x] = _percentile_sorted(x, ns); h},
    quartiles: [25, 50, 75].inject({}){|h,x| h[x] = _percentile_sorted(x, ns); h}
  }
end

# prng = Random.new
# puts stats((1..10000).map{|_| prng.rand(6)+1 + prng.rand(6)+1})
#
# puts stats((1..10000).map{|_| prng.rand(1.0)}, include_histogram=false)