gistfile1.py

from __future__ import print_function
from os.path import basename

import math
import sys

import numpy as np

# FIXME: look into decimal type

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]

    # Analaysis
    p = { key: None for key in [.50, .75, .85, .90, .95, .99] }

    with open(sys.argv[1], 'rU') as f:
        lines = map(float, f)
        arr = np.array(lines)
        count = arr.size
        maximum = np.amax(arr)

    # Presentation
    #print("Country\tp50\tp75\tp85\tp90\tp95\tp99\tMean\tMax\tSigma\tCount")

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

    # Percentiles
    for k in sorted(p.keys()):
        res = np.percentile(arr, k * 100)
        print("%d\t" % res, end='')

    # Misc stats
    print("%d\t%d\t%d\t%d" % (0, maximum, 0, count))

if __name__ == '__main__':
    main()