def test_fill_missing_dates_shifted_clean_config_int(self):
dates = pd.Series([
"2020/08/18 12:00",
"2020/08/18 12:16",
"2020/08/18 12:21",
"2020/08/18 12:31",
])
dates = pd.to_datetime(dates)
glucose = pd.Series([0, 1, 2, 3], dtype=np.float64)
df = pd.DataFrame({DT: dates, GLUCOSE: glucose})
fill_config = configs.CleanConfig(interval=5,
use_api=False,
fill_glucose_tolerance=2)
self.FileCleaner.set_clean_config(fill_config)
filled = self.FileCleaner._fill_missing_dates(df)
expect = pd.DataFrame({
DT:
pd.date_range("2020-08-18 12:00", periods=7, freq="5min"),
GLUCOSE: [0.0, np.nan, np.nan, 1.0, 2.0, np.nan, 3.0]
})
self.assertTrue(
filled.equals(expect),
"\nMethod result:\n{}\nExpected:\n{}\n".format(filled, expect))
def test_call_with_wrong_api_name(self):
wrong_api_config = configs.CleanConfig(interval=5,
use_api="wrong_api_name")
array = np.array(80 * [7])
fixer = DateFixer.DateFixer(wrong_api_config)
with self.assertRaises(ValueError):
fixer(array)
def test_predict_api_fallback_on_local(self, mocked_post, mocked_json,
mocked_raise):
api_config = configs.CleanConfig(interval=5, use_api="metronome")
fixer = DateFixer.DateFixer(api_config)
fixer._predict_local = mock.Mock()
example_dict = {"key": "test"}
fixer._predict_api(example_dict)
fixer._predict_local.assert_called_once()
def test_fix_dates_clean_config_int(self):
long_df = pd.DataFrame({
DT:
pd.date_range("2020/08/18 19:00", "2020/08/18 22:00", freq="5min"),
GLUCOSE:
37 * [80],
})
api_config = configs.CleanConfig(5, use_api="metronome")
self.FileCleaner.set_clean_config(api_config)
res = self.FileCleaner.fix_dates(long_df)
np.testing.assert_allclose(res, np.array(37 * [1]))
def test_predict_api_calls(self, mocked_post, mocked_json):
api_config = configs.CleanConfig(interval=5, use_api="metronome")
fixer = DateFixer.DateFixer(api_config)
mocked_post.raise_for_status = mock.Mock()
example_dict = {"key": "test"}
call_res = fixer._predict_api(example_dict)
self.assertTrue(call_res)
mocked_post.assert_called_once_with(
api_config._construct_full_api_address(),
json=json.dumps(example_dict),
timeout=0.5)
def test_fill_glucose_values_with_nearest_clean_config_int(self):
dates = pd.Series([
"2020/08/18 12:00",
"2020/08/18 12:01",
"2020/08/18 12:05",
"2020/08/18 12:10",
"2020/08/18 12:11",
"2020/08/18 12:15",
"2020/08/18 12:20",
"2020/08/18 12:25",
"2020/08/18 12:29",
"2020/08/18 12:30",
])
dates = pd.to_datetime(dates)
glucose = pd.Series([np.nan, 1, 2, np.nan, 4, 5, 6, 7, 8, np.nan],
dtype=np.float64)
not_fixed = pd.DataFrame({DT: dates, GLUCOSE: glucose})
not_fixed[GLUCOSE] = not_fixed[GLUCOSE].astype(np.float64)
date_fixed = not_fixed.iloc[[0, 2, 3, 5, 6, 7, 9], :].copy()
fill_config = configs.CleanConfig(interval=5,
use_api=False,
fill_glucose_tolerance=2)
self.FileCleaner.set_clean_config(fill_config)
filled = self.FileCleaner._fill_glucose_values_with_nearest(
date_fixed, not_fixed)
expect = pd.DataFrame(
{
DT: pd.date_range(
"2020-08-18 12:00:00", periods=7, freq="5min"),
GLUCOSE: np.array([1, 2, 4, 5, 6, 7, 8], dtype=np.float),
},
index=[0, 2, 3, 5, 6, 7, 9])
self.assertTrue(filled.equals(expect))
def setUp(self):
self.config_5 = configs.CleanConfig(interval=5, use_api=False)
self.api_config_5 = configs.CleanConfig(interval=5,
use_api="metronome")
self.fixer_5 = DateFixer.DateFixer(self.config_5)
self.variables_no = WINDOW_SIZE - 1
def test_tidy_simple_df_clean_config_int(self):
# DF setup
# couple of issues to solve in this dates
dates = [
"2020/08/18 12:00",
"2020/08/18 12:05",
"2020/08/18 12:05:01", # shift by 1sec
"2020/08/18 12:10:01",
"2020/08/18 12:15:01",
"2020/08/18 12:20:02", # another shift
"2020/08/18 12:25:02",
np.nan, # nan
"2020/08/18 12:30:02",
"2020/08/18 12:35:02",
"2020/08/18 12:35:57", # additional timepoints
"2020/08/18 12:37:00",
"2020/08/18 12:40:02",
"2020/08/18 12:45:02",
"2020/08/18 12:50:02",
"2020/08/18 12:55:02",
"2020/08/18 13:00:02",
"2020/08/18 13:05:02",
"2020/08/18 13:10:02",
"2020/08/18 13:15:01",
"2020/08/18 13:20:01",
"2020/08/18 13:25",
"2020/08/18 13:30",
"2020/08/18 13:35",
"2020/08/18 13:40",
"2020/08/18 13:50", # missing timepoint
"2020/08/18 13:55",
"2020/08/18 14:00",
"2020/08/18 14:05",
"2020/08/18 14:10",
"2020/08/18 14:15",
"2020/08/18 14:20",
"2020/08/18 14:25",
"2020/08/18 14:30",
"2020/08/18 14:35",
"2020/08/18 14:40",
]
dates = pd.to_datetime(dates)
glucose = np.array(len(dates) * [80], dtype=np.float)
glucose[9] = np.nan
df = pd.DataFrame({
DT: dates,
GLUCOSE: glucose,
})
# Clean config setup
clean_config = configs.CleanConfig(interval=5,
use_api="metronome",
fill_glucose_tolerance=2)
cleaner = FileCleaner(df, clean_config)
# Tidy!
res = cleaner.tidy()
# Expected setup
expect_index = pd.date_range("2020-08-18 12:00",
"2020-08-18 14:40",
freq="5min")
glucose = len(expect_index) * [80]
glucose[7] = np.nan
glucose[21] = np.nan
expect = pd.DataFrame({
DT: expect_index,
GLUCOSE: glucose,
})
self.assertTrue(res.equals(expect))