one_hot_svm.py

#!/usr/bin/env python2
# -*- coding: utf-8 -*-
"""
Created on Wed Jul  5 00:19:18 2017
@author: saliksyed
"""
import json
from sklearn.cluster import KMeans

country_routes = json.loads(open("routes.json").read())



countries_to_index = {}
idx = 0

for country in country_routes:
    for key in country_routes[country]["to"].keys():
        if not key in countries_to_index:
            countries_to_index[key] = idx
            idx += 1

total_countries = idx + 1

vectors = []
countries = []
for country in country_routes:
    vec = [0] * total_countries
    for key in country_routes[country]["to"].keys():
        vec[countries_to_index[key]] = 1
    
    vectors.append(vec)
    countries.append(country)


from sklearn import svm


country_to_country_code = json.loads(open("country_name_to_country_code.json").read())
populations = json.loads(open("population_by_country_code.json").read())
gdps = json.loads(open("gdp_by_country_code.json").read())

X = []
y = []

for i, country in enumerate(countries):
    if country in country_to_country_code:
        code = country_to_country_code[country]
        if populations[code] != None:
            X.append(vectors[i])
            if populations[code] > 1000000:
                y.append(1)
            else:
                y.append(0)

# now we'll keep most of the data is training data, and keep the rest to "test"
# if our algorithm actually works on new examples it hasn't seen before

num_train = int(len(X)/2.0)

X_train = X[:num_train]
y_train = y[:num_train]

X_test = X[num_train:]
y_test = y[num_train:]
    
clf = svm.SVC()
clf.fit(X_train, y_train)


predictions = clf.predict(X_test)
print predictions

# Now let's check our predictions:
correct = 0
for i, prediction in enumerate(predictions):
    if prediction == y_test[i]:
        correct += 1

print "Percentage correct:"
print float(correct) / len(predictions) * 100