stats-by-country.py

stats-by-country.py

#!/usr/bin/python3.4

import math
import statistics as stats
import signal
import sys

from os.path import basename

# Hide stack trace from KeyboardInterrupt
signal.signal(signal.SIGINT, lambda signal, frame: sys.exit(1))

def main():
    if len(sys.argv) != 2:
        sys.exit("%s: <infile>" % basename(sys.argv[0]))

    # Country codes conform to format defined by ISO 3166-1 alpha-2
    ccode = basename(sys.argv[1]).split('.')[0]

    percentile = {.25: '', .75: '', .85: '', .90: '', .95: '', .99: ''}

    with open(sys.argv[1], 'rU') as f:
        infile = sorted([int(line.split(',')[6].partition('.')[0])
                        for line in f])

    total = len(infile)
    maximum = infile[-1]

    # Header
    print("Country\tp25\tp75\tp85\tp90\tp95\tp99\tMean\tMedian\tMode\tMax\t"
          "Sigma\tCV\t%RSD\tIQR\tSamples")

    # Country
    print("%s\t" % (ccode), end='')

    # Percentiles
    for p in sorted(percentile.keys()):
        i = math.ceil(total * p)
        percentile[p] = infile[i]
        print("%d\t" % percentile.get(p), end='')

    # Misc stats
    try:
        mode = stats.mode(infile)
    except stats.StatisticsError:
        mode = 'x'

    mean   = stats.mean(infile)
    median = stats.median(infile)
    sigma  = stats.stdev(infile)
    cv     = (sigma/mean)
    rsd    = ((sigma/mean) * 100)
    iqr    = ((percentile.get(.75) - percentile.get(.25)) / median)

    print("%d\t%d\t%s\t%d\t%d\t%0.4f\t%d%%\t%0.4f\t%d" %
         (mean, median, mode, maximum, sigma, cv, rsd, iqr, total))

if __name__ == '__main__':
    main()