def profile_lending_club():
import datetime
from uuid import uuid4
now = datetime.datetime.utcnow()
session_id = uuid4().hex
df = pd.read_csv(
os.path.join(_MY_DIR, os.pardir, "testdata", "lending_club_1000.csv"))
profile = DatasetProfile(name="test",
session_id=session_id,
session_timestamp=now)
profile.track_dataframe(df)
return profile
def test_dataframe_profile():
time = datetime.datetime.now()
df = util.testing.makeDataFrame()
profile = DatasetProfile("test", time)
profile.track_dataframe(df)
profile_factory = dataframe_profile(df, name="test", timestamp=time)
assert profile_factory.columns["A"].number_tracker.variance.mean == profile.columns["A"].number_tracker.variance.mean
profile_factory_2 = dataframe_profile(df)
assert profile_factory_2.columns["A"].number_tracker.variance.mean == profile.columns["A"].number_tracker.variance.mean
profile_factory_3 = dataframe_profile(df, timestamp=103433)
assert profile_factory_3.columns["A"].number_tracker.variance.mean == profile.columns["A"].number_tracker.variance.mean
