geocorrelation.py

import numpy as np
import re  # regular expressions
import csv
from matplotlib import pyplot as pl
from scipy.stats.stats import pearsonr


def parse_dms(dms):
    parts = re.split('[^\d\w]+', dms)
    lat = dms2dd(parts[0], parts[1], parts[2])
    # lng = dms2dd(parts[4], parts[5], parts[6], parts[7])

    return lat


def dms2dd(degrees, minutes, direction):
    dd = float(degrees) + float(minutes) / 60
    if direction == 'S' or direction == 'W':
        dd *= -1
    return dd

# Create dictionaries of geolocation:
with open('geoposition.csv', 'rU') as geofile:
    georeader = csv.reader(geofile, delimiter=',')
    next(georeader, None)  # skip header
    latitudes = {row[0]: parse_dms(row[1]) for row in georeader}

with open('geoposition.csv', 'rU') as geofile:
    georeader = csv.reader(geofile, delimiter=',')
    next(georeader, None)  # skip header
    longitudes = {row[0]: parse_dms(row[2]) for row in georeader}

# Create dictionary of roundedness:
with open('roundedness.csv', 'rU') as geofile:
    roundreader = csv.reader(geofile, delimiter=',')
    next(roundreader, None)  # skip header
    roundedness = {row[0]: float(row[1]) for row in roundreader}


xround = []
ylatit = []
ylongi = []
# Create lists to determine correlation:
for index, roundkey in enumerate(roundedness):
    for geokey in latitudes:
        if roundkey in geokey:
            xround.append(roundedness[roundkey])
            ylatit.append(latitudes[geokey])
            ylongi.append(longitudes[geokey])

fig = pl.figure()
pl.scatter(np.absolute(ylongi), xround)
trendline = np.poly1d(np.polyfit(np.absolute(ylongi), xround, 1))
pl.plot(np.absolute(ylongi), trendline(np.absolute(ylongi)), 'r--')
pl.xlabel('Longitude of Capital')
pl.ylabel('Roundness of Country')
print(pearsonr(np.absolute(ylongi), xround))
pl.show()

geoposition.csv

roundedness.csv