def create_dha(path='data/dha.csv'): test_tablename = 'dhatest' + str(int(time.time() * 1000000)) + str(int(random.random()*10000000)) header, rows = data_utils.read_csv(path) create_btable_result = engine.create_btable(test_tablename, header, rows, key_column=0) metadata = engine.persistence_layer.get_metadata(test_tablename) global test_tablenames test_tablenames.append(test_tablename) return test_tablename, create_btable_result
def create_dha(path='data/dha.csv'):
test_tablename = 'dhatest' + str(int(time.time() * 1000000)) + \
str(int(random.random()*10000000))
header, rows = data_utils.read_csv(path)
create_btable_result = engine.create_btable(test_tablename, header, rows, key_column=0)
# metadata = engine.persistence_layer.get_metadata(test_tablename)
global test_tablenames
test_tablenames.append(test_tablename)
return test_tablename, create_btable_result
def create_describe_btable(data_path='data/describe.csv',
codebook_path='data/describe_codebook.csv', use_codebook=True):
# TODO: refactor codebook generation to Engine, not Client
test_tablename = 'describetest' + str(int(time.time() * 1000000)) + \
str(int(random.random()*10000000))
if use_codebook:
codebook_header, codebook_rows = data_utils.read_csv(codebook_path)
codebook = dict()
for codebook_row in codebook_rows:
codebook[codebook_row[0]] = dict(zip(['short_name', 'description', 'value_map'],
codebook_row[1:]))
else:
codebook = None
header, rows = data_utils.read_csv(data_path)
create_btable_result = engine.create_btable(test_tablename, header, rows, key_column=0,
codebook=codebook)
global test_tablenames
test_tablenames.append(test_tablename)
return test_tablename, create_btable_result
def create_describe_btable(data_path='data/describe.csv',
codebook_path='data/describe_codebook.csv', use_codebook=True):
# TODO: refactor codebook generation to Engine, not Client
test_tablename = 'describetest' + str(int(time.time() * 1000000)) + \
str(int(random.random()*10000000))
if use_codebook:
codebook_header, codebook_rows = data_utils.read_csv(codebook_path)
codebook = dict()
for codebook_row in codebook_rows:
codebook[codebook_row[0]] = dict(zip(['short_name', 'description', 'value_map'],
codebook_row[1:]))
else:
codebook = None
header, rows = data_utils.read_csv(data_path)
create_btable_result = engine.create_btable(test_tablename, header, rows, key_column=0,
codebook=codebook)
global test_tablenames
test_tablenames.append(test_tablename)
return test_tablename, create_btable_result
def test_subsampling():
# Use Kiva table, which has 10000 rows, instead of DHA.
test_tablename = 'kivatest' + str(int(time.time() * 1000000)) + \
str(int(random.random()*10000000))
global test_tablenames
test_tablenames.append(test_tablename)
path = 'data/kiva_small.csv'
header, rows = data_utils.read_csv(path)
num_rows = 4 # rows in kiva_small
num_rows_subsample = 2
# client('create btable %s from %s' % (test_tablename, path), debug=True, pretty=False)
# only analyze using some rows
engine.create_btable(test_tablename, header, rows, subsample=num_rows_subsample, key_column=0)
# make sure select (using no models) works and returns the correct number of rows
functions = bql.bql_statement.parseString('select loan_id, loan_status from test',
parseAll=True).functions
whereclause = None
limit = float('inf')
order_by = False
select_result = engine.select(test_tablename, functions, whereclause, limit, order_by, None)
assert len(select_result['data']) == num_rows # number of rows in Kiva
# TODO: better testing to see what we can do before subsampling (with partial models)
num_models = 2
iterations = 1
engine.initialize_models(test_tablename, num_models)
# analyze segfaults
engine.analyze(test_tablename, model_indices='all', iterations=iterations, background=False)
print('analyzed')
model_ids = engine.persistence_layer.get_model_ids(test_tablename)
for i in range(num_models):
model = engine.persistence_layer.get_models(test_tablename, i)
assert model['iterations'] == iterations
# make sure normal queries work and return the correct number of rows
functions = bql.bql_statement.parseString('select loan_id, predictive probability of '
'loan_status from test', parseAll=True).functions
whereclause = None
limit = float('inf')
order_by = False
select_result = engine.select(test_tablename, functions, whereclause, limit, order_by, None)
assert len(select_result['data']) == num_rows # number of rows in Kiva
def test_subsampling():
# Use Kiva table, which has 10000 rows, instead of DHA.
test_tablename = 'kivatest' + str(int(time.time() * 1000000)) + \
str(int(random.random()*10000000))
global test_tablenames
test_tablenames.append(test_tablename)
path = 'data/kiva_small.csv'
header, rows = data_utils.read_csv(path)
num_rows = 4 # rows in kiva_small
num_rows_subsample = 2
# client('create btable %s from %s' % (test_tablename, path), debug=True, pretty=False)
# only analyze using some rows
engine.create_btable(test_tablename, header, rows, subsample=num_rows_subsample, key_column=0)
# make sure select (using no models) works and returns the correct number of rows
functions = bql.bql_statement.parseString('select loan_id, loan_status from test',
parseAll=True).functions
whereclause = None
limit = float('inf')
order_by = False
select_result = engine.select(test_tablename, functions, whereclause, limit, order_by, None)
assert len(select_result['data']) == num_rows # number of rows in Kiva
# TODO: better testing to see what we can do before subsampling (with partial models)
num_models = 2
iterations = 1
engine.initialize_models(test_tablename, num_models)
# analyze segfaults
engine.analyze(test_tablename, model_indices='all', iterations=iterations, background=False)
print('analyzed')
model_ids = engine.persistence_layer.get_model_ids(test_tablename)
for i in range(num_models):
model = engine.persistence_layer.get_models(test_tablename, i)
assert model['iterations'] == iterations
# make sure normal queries work and return the correct number of rows
functions = bql.bql_statement.parseString('select loan_id, predictive probability of '
'loan_status from test', parseAll=True).functions
whereclause = None
limit = float('inf')
order_by = False
select_result = engine.select(test_tablename, functions, whereclause, limit, order_by, None)
assert len(select_result['data']) == num_rows # number of rows in Kiva
