is-baidu-world-cup-prediction-reliable

analyse.py

import json

f = open('all.json')
json = json.loads(f.read())
f.close()

predicts = []
results = []
for i in json:
    bodys = i['data']['body']
    for body in bodys:
        predict = [int(body['win_rate']), int(body['equal_rate']), int(body['fail_rate'])]
        predicts.append(predict)
        if int(body['home_score']) > int(body['away_score']):
            result = 1
        elif int(body['home_score']) == int(body['away_score']):
            result = 0
        else:
            result = -1
        results.append(result)

def baidu_predict(index):
    return predicts[index];

def absolute_paul_predict(index):
    if results[index] == 1:
        return [100, 0, 0]
    elif results[index] == 0:
        return [0, 100, 0]
    else:
        return [0, 0, 100]

def monkey_predict(index):
    return [33.33, 33.33, 33.33]

def gambler_predict(index):
    return [50, 0, 50]

def test_predict(predict_func):
    score = 0
    for i in range(len(results)):
        score += predict_func(i)[1 - results[i]]
    return score

print len(results)

print test_predict(baidu_predict)
print test_predict(absolute_paul_predict)
print test_predict(monkey_predict)
print test_predict(gambler_predict)

concat.py

import os.path
import json
time = 1402588800
fname = str(time) + '.json'
jsons = []
while os.path.isfile(fname):
    f = open(fname, 'r')
    _json = json.loads(f.read())
    f.close()
    jsons.append(_json)
    time += 60 * 60 * 24
    fname = str(time) + '.json'

f = open('all.json', 'w')
f.write(json.dumps(jsons))
f.close()

task.sh

curl http://trends.baidu.com/worldcup/events/ajax/time?query=1402588800 > 1402588800.json
curl http://trends.baidu.com/worldcup/events/ajax/time?query=1402675200 > 1402675200.json
curl http://trends.baidu.com/worldcup/events/ajax/time?query=1402761600 > 1402761600.json
curl http://trends.baidu.com/worldcup/events/ajax/time?query=1402848000 > 1402848000.json
curl http://trends.baidu.com/worldcup/events/ajax/time?query=1402934400 > 1402934400.json
curl http://trends.baidu.com/worldcup/events/ajax/time?query=1403020800 > 1403020800.json
curl http://trends.baidu.com/worldcup/events/ajax/time?query=1403107200 > 1403107200.json
curl http://trends.baidu.com/worldcup/events/ajax/time?query=1403193600 > 1403193600.json
curl http://trends.baidu.com/worldcup/events/ajax/time?query=1403280000 > 1403280000.json
curl http://trends.baidu.com/worldcup/events/ajax/time?query=1403366400 > 1403366400.json
curl http://trends.baidu.com/worldcup/events/ajax/time?query=1403452800 > 1403452800.json

tasks.py

f = open('task.sh', 'w');

time = 1402588800
cmd_pattern = "curl http://trends.baidu.com/worldcup/events/ajax/time?query=%d > %d.json"
for i in range(13, 23 + 1):
    f.write(cmd_pattern % (time, time) + "\n")
    time += 60 * 60 * 24