def test_load(self):
print "TESTING: load data"
result = Result()
data = result.load_results(self.jsonfile)
self.assertTrue(isinstance(data,pandas.DataFrame))
self.assertTrue(data.shape[0] == 44)
self.assertTrue(data.shape[1] == 13)
def setUp(self):
self.pwd = get_installdir()
self.tmpdir = tempfile.mkdtemp()
self.jsonfile = os.path.abspath("%s/tests/data/results/results.json" %
self.pwd)
self.result = Result()
self.result.load_results(self.jsonfile)
def test_load(self):
print "TESTING: load data"
result = Result()
data = result.load_results(self.jsonfile)
self.assertTrue(isinstance(data, pandas.DataFrame))
self.assertTrue(data.shape[0] == 44)
self.assertTrue(data.shape[1] == 13)
def download_data(data_loc,
access_token=None,
filters=None,
battery=None,
save=True,
url=None,
file_name=None):
start_time = time()
#Load Results from Database
results = Result(access_token, filters=filters, url=url)
data = results.data
if 'experiment_exp_id' not in data.columns:
data.loc[:, 'experiment_exp_id'] = [
x['exp_id'] for x in data['experiment']
]
if 'experiment_template' not in data.columns:
data.loc[:, 'experiment_template'] = [
x['template'] for x in data['experiment']
]
if battery:
data = result_filter(data, battery=battery)
# remove duplicates
remove_duplicates(data)
# remove a few mistakes from data
data = data.query('worker_id not in ["A254JKSDNE44AM", "A1O51P5O9MC5LX"]'
) # Sandbox workers
data.reset_index(drop=True, inplace=True)
# if saving, save the data and the lookup file for anonymized workers
if save == True:
if file_name == None:
file_name = 'mturk_data.json'
if file_name[-4:] == 'json':
data.to_json(os.path.join(data_loc, file_name))
elif file_name[-3:] == 'pkl':
data.to_pickle(os.path.join(data_loc, file_name))
print('Finished saving')
finish_time = (time() - start_time) / 60
print('Finished downloading data. Time taken: ' + str(finish_time))
return data
for col in drop_columns:
filters[col] = {'drop': True}
# Strip token from specified file
f = open(token)
access_token = f.read().strip()
# Set up variables for the download request
battery = 'Self Regulation Retest Battery'
url = 'http://www.expfactory.org/new_api/results/62/'
file_name = 'mturk_retest_data.json'
fields = get_result_fields()
# Create results object
results = Result(access_token, filters=filters, url=url)
# Clean filters from results objects
results.clean_results(filters)
# Extract data from the results object
data = results.data
# Remainder of download_data
data = result_filter(data, battery=battery)
remove_duplicates(data)
data = data.query(
'worker_id not in ["A254JKSDNE44AM", "A1O51P5O9MC5LX"]') # Sandbox workers
data.reset_index(drop=True, inplace=True)
# Save data
from expanalysis.experiments.utils import remove_duplicates from expanalysis.results import get_filters, Result # default filters to clean up the downloaded results object filters = get_filters() # expfactory.org/token access_token = "1111111111" # expfactory.org/token # url of the battery # for example, if the battery url is: http://expfactory.org/batteries/999/ # the url below will be: http://www.expfactory.org/new_api/results/999/ url = 'http://www.expfactory.org/new_api/results/999/' # create a results object results = Result(access_token, filters=filters, url=url) # we care about the data data = results.data # remove duplicates - there shouldn't be any, just a safety precaution remove_duplicates(data) """ The results.data object has one row per worker/experiment pair. The column "data" holds the data as a dictionary. Unofrunately, this data is in a 1-length list, and the data's "trialdata" is really what you want. To index the first row's data you would index the results.data object like so: >>> first_row_data = results.data.iloc[0]['data'][0]['trialdata'] Which can then be converted into a dataframe easily >>> first_row_data = pandas.DataFrame(first_row_data) """
class TestAPI(unittest.TestCase):
def setUp(self):
self.pwd = get_installdir()
self.tmpdir = tempfile.mkdtemp()
self.jsonfile = os.path.abspath("%s/tests/data/results/results.json" %self.pwd)
self.result = Result()
self.result.load_results(self.jsonfile)
def tearDown(self):
shutil.rmtree(self.tmpdir)
def test_check_numeric(self):
not_numeric = ["hello","goodbye"]
numeric_float = [1.1,2.2,3.3]
numeric_int = [1,2,3]
not_numeric_mixed = ["hello",2,3.0]
self.assertTrue(check_numeric(numeric_float))
self.assertTrue(check_numeric(numeric_int))
self.assertTrue(check_numeric(not_numeric)==False)
self.assertTrue(check_numeric(not_numeric_mixed)==False)
def test_filter(self):
filtered = self.result.filter(field="experiment_exp_id",value="bridge_game")
self.assertTrue(filtered.shape[0]==20)
self.assertTrue(len(numpy.unique(filtered["experiment_exp_id"]))==1)
def test_load(self):
print "TESTING: load data"
result = Result()
data = result.load_results(self.jsonfile)
self.assertTrue(isinstance(data,pandas.DataFrame))
self.assertTrue(data.shape[0] == 44)
self.assertTrue(data.shape[1] == 13)
def test_experiment_extract(self):
print "TESTING: experiment extraction"
experiment = self.result.extract_experiment(exp_id="stroop")
experiment_columns = [u'block_duration', u'condition', u'correct', u'correct_response',
u'current_trial', u'dateTime', u'feedback_duration',
u'internal_node_id', u'key_press', u'possible_responses', u'responses',
u'rt', u'stim_color', u'stim_duration', u'stim_word', u'stimulus',
u'time_elapsed', u'timing_post_trial', u'trial_id', u'trial_index',
u'trial_type', u'trialdata', u'uniqueid', u'view_history']
self.assertTrue(isinstance(experiment,pandas.DataFrame))
self.assertTrue(experiment.shape[0]==747)
[self.assertTrue(x) in experiment.columns for x in experiment_columns]
def test_survey_extract(self):
print "TESTING: survey extraction"
survey = self.result.extract_experiment(exp_id="bis11_survey")
self.assertTrue(isinstance(survey,pandas.DataFrame))
def test_game_extract(self):
print "TESTING: game extraction"
game = self.result.extract_experiment(exp_id="bridge_game")
game_columns = [u'current_trial', u'uniqueid', u'dateTime', u'ACC', u'RT', u'solution',
u'problem_id', u'trial', u'finished', u'points', u'answer', u'n1',
u'n2', u'problem']
self.assertTrue(isinstance(game,pandas.DataFrame))
self.assertTrue(game.shape[0]==301)
[self.assertTrue(x) in game.columns for x in game_columns]
def setUp(self):
self.pwd = get_installdir()
self.tmpdir = tempfile.mkdtemp()
self.jsonfile = os.path.abspath("%s/tests/data/results/results.json" % self.pwd)
self.result = Result()
self.result.load_results(self.jsonfile)
class TestAPI(unittest.TestCase):
def setUp(self):
self.pwd = get_installdir()
self.tmpdir = tempfile.mkdtemp()
self.jsonfile = os.path.abspath("%s/tests/data/results/results.json" % self.pwd)
self.result = Result()
self.result.load_results(self.jsonfile)
def tearDown(self):
shutil.rmtree(self.tmpdir)
def test_check_numeric(self):
not_numeric = ["hello", "goodbye"]
numeric_float = [1.1, 2.2, 3.3]
numeric_int = [1, 2, 3]
not_numeric_mixed = ["hello", 2, 3.0]
self.assertTrue(check_numeric(numeric_float))
self.assertTrue(check_numeric(numeric_int))
self.assertTrue(check_numeric(not_numeric) == False)
self.assertTrue(check_numeric(not_numeric_mixed) == False)
def test_filter(self):
filtered = self.result.filter(field="experiment_exp_id", value="bridge_game")
self.assertTrue(filtered.shape[0] == 20)
self.assertTrue(len(numpy.unique(filtered["experiment_exp_id"])) == 1)
def test_load(self):
print "TESTING: load data"
result = Result()
data = result.load_results(self.jsonfile)
self.assertTrue(isinstance(data, pandas.DataFrame))
self.assertTrue(data.shape[0] == 44)
self.assertTrue(data.shape[1] == 13)
def test_experiment_extract(self):
print "TESTING: experiment extraction"
experiment = self.result.extract_experiment(exp_id="stroop")
experiment_columns = [
u"block_duration",
u"condition",
u"correct",
u"correct_response",
u"current_trial",
u"dateTime",
u"feedback_duration",
u"internal_node_id",
u"key_press",
u"possible_responses",
u"responses",
u"rt",
u"stim_color",
u"stim_duration",
u"stim_word",
u"stimulus",
u"time_elapsed",
u"timing_post_trial",
u"trial_id",
u"trial_index",
u"trial_type",
u"trialdata",
u"uniqueid",
u"view_history",
]
self.assertTrue(isinstance(experiment, pandas.DataFrame))
self.assertTrue(experiment.shape[0] == 747)
[self.assertTrue(x) in experiment.columns for x in experiment_columns]
def test_survey_extract(self):
print "TESTING: survey extraction"
survey = self.result.extract_experiment(exp_id="bis11_survey")
self.assertTrue(isinstance(survey, pandas.DataFrame))
def test_game_extract(self):
print "TESTING: game extraction"
game = self.result.extract_experiment(exp_id="bridge_game")
game_columns = [
u"current_trial",
u"uniqueid",
u"dateTime",
u"ACC",
u"RT",
u"solution",
u"problem_id",
u"trial",
u"finished",
u"points",
u"answer",
u"n1",
u"n2",
u"problem",
]
self.assertTrue(isinstance(game, pandas.DataFrame))
self.assertTrue(game.shape[0] == 301)
[self.assertTrue(x) in game.columns for x in game_columns]
for col in drop_columns:
filters[col] = {'drop': True}
# Strip token from specified file
f = open(token)
access_token = f.read().strip()
# Set up variables for the download request
battery = 'Self Regulation Retest Battery'
url = 'http://www.expfactory.org/new_api/results/62/'
file_name = 'mturk_retest_data.json'
fields = get_result_fields()
# Create results object
results = Result(access_token, filters = filters, url = url)
# Clean filters from results objects
results.clean_results(filters)
# Extract data from the results object
data = results.data
# Remainder of download_data
data = result_filter(data, battery = battery)
remove_duplicates(data)
data = data.query('worker_id not in ["A254JKSDNE44AM", "A1O51P5O9MC5LX"]') # Sandbox workers
data.reset_index(drop = True, inplace = True)
# Save data
data.to_json(path.join(data_dir, file_name))
