def testInit(self):
print 'testInit ...'
# Test with an invalid config file.
with self.assertRaises(TypeError):
trials = Trials(None, True)
with self.assertRaises(IOError):
trials = Trials('Nonexistent file', True)
# Create a config file.
mmxConfig = MmxConfig()
presFile = 'tests/WLBG-geog.csv'
startDate = '1-1-2016'
endDate = '1-1-2017'
outDir = '../'
numProcs = 500
numTrials = 200
species = 'WLBG'
mmxConfig.initializeFromValues(presFile, startDate, endDate, species,
outDir, numProcs, numTrials)
mmxConfig.write()
# Initialize GetMerra
mmxConfig.setStateComplete()
trials = Trials(mmxConfig.getConfigFile(), True)
def main():
conf = Config()
db = Database(conf.address, conf.username, conf.password,
conf.database_name)
valid_studies_counter = 0
for i in range(0, count):
try:
scraped_url = "http://drks-neu.uniklinik-freiburg.de/drks_web/navigate.do?navigationId=trial.HTML&TRIAL_ID=DRKS" + str(
i).zfill(8)
request = requests.get(scraped_url, stream=True)
request.encoding = 'utf-8'
my_scraper = BeautifulSoup(request.text.encode('utf8'), "lxml")
if not (my_scraper.find('ul', class_="error")
or my_scraper.find('ul', class_="errors")):
# if true webpage contains valid data of study
Trials(my_scraper, db)
valid_studies_counter += 1
sys.stdout.write("\r Processing " + str(i) + "/" + str(count) +
" => " + str(i * 100 / count) + " percent\t\t" +
str(valid_studies_counter) + " studies found")
sys.stdout.flush()
except AttributeError:
print("\n[-] attribute error at " + str(i))
except Exception:
print("\n[-] unknown error at " + str(i))
db.close_connection()
def testGenerateFileIndexes(self):
print 'testGenerateFileIndexes ...'
mmxConfig = MmxConfig()
presFile = 'tests/WLBG-geog.csv'
startDate = '1-1-2016'
endDate = '1-1-2017'
outDir = '../'
numProcs = 500
numTrials = 200
species = 'WLBG'
mmxConfig.initializeFromValues(presFile, startDate, endDate, species,
outDir, numProcs, numTrials)
mmxConfig.write()
trials = Trials(mmxConfig.getConfigFile(), True)
trials.generateFileIndexes(50)
def testGetPhase(self):
print 'testGetPhase ...'
mmxConfig = MmxConfig()
presFile = 'tests/WLBG-geog.csv'
startDate = '1-1-2016'
endDate = '1-1-2017'
outDir = '../'
numProcs = 500
numTrials = 200
species = 'WLBG'
mmxConfig.initializeFromValues(presFile, startDate, endDate, species,
outDir, numProcs, numTrials)
mmxConfig.write()
# Initialize GetMerra
mmxConfig.setStateComplete()
trials = Trials(mmxConfig.getConfigFile(), True)
self.assertEqual(trials.config.phase, 'TRIALS')
def test(self, data, control, variations):
update = {}
buckets = [control] + variations
for bucket in buckets:
total = data[bucket][0]['count']
successes = data[bucket][1]['count']
update[bucket] = (successes, total - successes)
test = Trials(buckets)
test.update(update)
dominances = test.evaluate('dominance', control=control)
lifts = test.evaluate('expected lift', control=control)
intervals = test.evaluate('lift CI', control=control, level=95)
for variation in variations:
print('Variation {name}:'.format(name=variation))
s, t = update[variation]
print('* Successes: {0}, Total: {1}, Percent: {2:2.3f}%'.format(s, t, 100*float(s)/t))
print('* E[lift] = {value:.2%}'.format(value=lifts[variation]))
print('* P({lower:.2%} < lift < {upper:.2%}) = 95%'.format(lower=intervals[variation][0], upper=intervals[variation][2]))
print('* P({name} > {control}) = {value:.2%}'.format(name=variation, control=control, value=dominances[variation]))
print
"""Sociological poll example."""
import sys
sys.path.append('..')
from trials import Trials
if __name__ == '__main__':
test = Trials(['Poroshenko', 'Tymoshenko'])
test.update({'Poroshenko': (48, 52), 'Tymoshenko': (12, 88)})
estimates = test.evaluate('posterior CI')
dominance = test.evaluate('dominance', control='Tymoshenko')
print('Poroshenko estimated vote share: {lower:.2%} - {upper:.2%} '
'(95% credibility)'.format(lower=estimates['Poroshenko'][0],
upper=estimates['Poroshenko'][2]))
print('Tymoshenko estimated vote share: {lower:.2%} - {upper:.2%} '
'(95% credibility)'.format(lower=estimates['Tymoshenko'][0],
upper=estimates['Tymoshenko'][2]))
print('Chance that Poroshenko beats Tymoshenko based on the poll data: '
'{chance:.2%}'.format(chance=dominance['Poroshenko']))
import sys
sys.path.append('..')
from trials import Trials
if __name__ == '__main__':
test = Trials(['A', 'B', 'C'])
test.update({
'A': (50, 10),
'B': (75, 15),
'C': (20, 15)
})
lift = test.evaluate('lift', control='A')
domination = test.evaluate('domination', control='A')
for variation in ['B', 'C']:
print('Variation {name}:'.format(name=variation))
print('* lift = {value:.2%}'.format(value=lift[variation]))
print('* P({name} > {control}) = {value:.2%}' \
.format(name=variation, control='A', value=domination[variation]))
"""Sociological poll example."""
import sys
sys.path.append('..')
from trials import Trials
if __name__ == '__main__':
test = Trials(['Poroshenko', 'Tymoshenko'])
test.update({
'Poroshenko': (48, 52),
'Tymoshenko': (12, 88)
})
estimates = test.evaluate('posterior CI')
dominance = test.evaluate('dominance', control='Tymoshenko')
print('Poroshenko estimated vote share: {lower:.2%} - {upper:.2%} '
'(95% credibility)'
.format(lower=estimates['Poroshenko'][0],
upper=estimates['Poroshenko'][2]))
print('Tymoshenko estimated vote share: {lower:.2%} - {upper:.2%} '
'(95% credibility)'
.format(lower=estimates['Tymoshenko'][0],
upper=estimates['Tymoshenko'][2]))
print('Chance that Poroshenko beats Tymoshenko based on the poll data: '
'{chance:.2%}'.format(chance=dominance['Poroshenko']))
"""Simple A/B test example."""
import sys
sys.path.append('..')
from trials import Trials
if __name__ == '__main__':
test = Trials(['A', 'B', 'C'])
test.update({
'A': (50, 10),
'B': (75, 15),
'C': (20, 15)
})
dominance = test.evaluate('dominance', control='A')
lift = test.evaluate('expected lift', control='A')
interval = test.evaluate('lift CI', control='A', level=95)
for variation in ['B', 'C']:
print('Variation {name}:'.format(name=variation))
print('* E[lift] = {value:.2%}'.format(value=lift[variation]))
print('* P({lower:.2%} < lift < {upper:.2%}) = 95%'
.format(lower=interval[variation][0],
upper=interval[variation][2]))
print('* P({name} > {control}) = {value:.2%}'
.format(name=variation, control='A',
value=dominance[variation]))
