def test_compute_regression_sensor_intercept(self):
"""Verified with lm(y~x)."""
test_covariate = LocationSeries(
data={
date(2020, 1, 1): 1,
date(2020, 1, 2): 3,
date(2020, 1, 3): 5,
date(2020, 1, 4): 6,
date(2020, 1, 5): 7,
date(2020, 1, 6): 9,
date(2020, 1, 7): 12
})
test_response = LocationSeries(
data={
date(2020, 1, 1): 10,
date(2020, 1, 2): 16,
date(2020, 1, 3): 22,
date(2020, 1, 4): 29,
date(2020, 1, 5): 28,
date(2020, 1, 6): 35,
date(2020, 1, 7): 42
})
assert np.isclose(
compute_regression_sensor(date(2020, 1, 6), test_covariate,
test_response, True),
6.586207 + 3.275862 * 9)
def test_compute_regression_sensor_insufficient_data(self):
test_covariate = LocationSeries(
data={
date(2020, 1, 1): 1,
date(2020, 1, 2): 3,
date(2020, 1, 3): np.nan,
date(2020, 1, 4): 6,
date(2020, 1, 5): 7,
date(2020, 1, 6): 9,
date(2020, 1, 7): 12
})
test_response = LocationSeries(
data={
date(2020, 1, 1): 10,
date(2020, 1, 2): 16,
date(2020, 1, 3): 22,
date(2020, 1, 4): 29,
date(2020, 1, 5): 28,
date(2020, 1, 6): 35,
date(2020, 1, 7): 42
})
assert np.isnan(
compute_regression_sensor(date(2020, 1, 1), test_covariate,
test_response, False))
assert np.isnan(
compute_regression_sensor(date(2020, 1, 6), test_covariate,
test_response, False))
def test_add_data(self):
test_ls = LocationSeries(data={date(2020, 1, 1): 2})
test_ls.add_data(date(2020, 1, 3), 4)
assert test_ls == LocationSeries(data={
date(2020, 1, 1): 2,
date(2020, 1, 3): 4
})
def test_compute_sensors_covariates(self, mock_get_indicator_data,
mock_compute_ar_sensor,
mock_compute_regression_sensor):
"""Test ground truth sensor and non-na regression sensor ar returned"""
mock_get_indicator_data.return_value = {
("a", "b", "state", "ca"): ["placeholder"],
("x", "y", "state", "ca"): ["placeholder"]
}
mock_compute_ar_sensor.return_value = 1.5
mock_compute_regression_sensor.side_effect = [
2.5, np.nan
] # nan means 2nd sensor is skipped
test_sensors = [
SensorConfig("a", "b", "c", 1),
SensorConfig("x", "y", "z", 2)
]
test_ground_truth_sensor = SensorConfig("i", "j", "k", 3)
test_ground_truth = [LocationSeries("ca", "state")]
assert compute_sensors(
date(2020, 5, 5), test_sensors, test_ground_truth_sensor,
test_ground_truth, False) == {
SensorConfig("i", "j", "k", 3):
[LocationSeries("ca", "state", {date(2020, 5, 2): 1.5})],
SensorConfig("a", "b", "c", 1):
[LocationSeries("ca", "state", {date(2020, 5, 4): 2.5})],
}
def test_get_data_range_mean_impute(self):
test_ls = LocationSeries(data={
date(2020, 1, 1): 7,
date(2020, 1, 2): np.nan,
date(2020, 1, 3): 9
})
assert test_ls.get_data_range(date(2020, 1, 1), date(2020, 1, 3),
"mean") == [7, 8.0, 9]
assert test_ls.get_data_range(date(2020, 1, 1), date(2020, 1, 2),
"mean") == [7, 7]
def test_get_data_range_no_impute(self):
test_ls = LocationSeries(data={
date(2020, 1, 1): 7,
date(2020, 1, 2): np.nan,
date(2020, 1, 3): 9
})
assert test_ls.get_data_range(date(2020, 1, 1), date(2020, 1, 3),
None) == [7, np.nan, 9]
assert test_ls.get_data_range(date(2020, 1, 1), date(2020, 1, 2),
None) == [7, np.nan]
def test_historical_sensors_no_data(self, mock_historical):
"""Test nothing returned for any sensor."""
mock_historical.return_value = (LocationSeries(), [])
test_sensors = [
SensorConfig("i", "j", "k", 3),
SensorConfig("a", "b", "c", 1),
SensorConfig("x", "y", "z", 2)
]
test_ground_truth = [LocationSeries("ca", "state")]
assert historical_sensors(None, None, test_sensors,
test_ground_truth) == {}
def test_get_data_range_invalid_impute(self):
test_ls = LocationSeries(data={
date(2020, 1, 1): 7,
date(2020, 1, 2): np.nan,
date(2020, 1, 3): 9
})
with pytest.raises(
ValueError,
match="Invalid imputation method. Must be None or 'mean'"):
test_ls.get_data_range(date(2020, 1, 1), date(2020, 1, 3),
"fakeimpute")
def test_compute_ar_sensor_regularize(self, random_normal):
"""coefficients verified with lm.ridge(y~x1+x2, lambda=1*12/11)
x1 and x2 constructed by hand, lambda is scaled since lm.ridge does some scaling by n/(n-1)
"""
random_normal.return_value = 0
values = LocationSeries(
data={
date(2020, 1, 1): -4.27815483,
date(2020, 1, 2): -4.83962077,
date(2020, 1, 3): -4.09548122,
date(2020, 1, 4): -3.86647783,
date(2020, 1, 5): -2.64494168,
date(2020, 1, 6): -3.99573135,
date(2020, 1, 7): -3.4824841,
date(2020, 1, 8): -2.77490127,
date(2020, 1, 9): -3.64162355,
date(2020, 1, 10): -2.5762891,
date(2020, 1, 11): -2.46793048,
date(2020, 1, 12): -3.20454941,
date(2020, 1, 13): -1.77057154,
date(2020, 1, 14): -0.02058535,
date(2020, 1, 15): 0.81182691,
date(2020, 1, 16): 0.32741982
})
assert np.isclose(
compute_ar_sensor(date(2020, 1, 15), values, 2, 1),
-2.8784639 + 0.2315984 * (-1.77057154 - -3.48901547) / 0.7637391 +
0.5143709 * (-0.02058535 - -3.28005019) / 0.8645852)
def test__export_to_csv(self):
"""Test export creates the right file and right contents."""
test_sensor = SensorConfig(source="src",
signal="sig",
name="test",
lag=4)
test_value = LocationSeries("ca", "state", {date(2020, 1, 1): 1.5})
with tempfile.TemporaryDirectory() as tmpdir:
out_files = _export_to_csv(test_value,
test_sensor,
date(2020, 1, 5),
receiving_dir=tmpdir)
assert len(out_files) == 1
out_file = out_files[0]
assert os.path.isfile(out_file)
assert out_file.endswith(
"issue_20200105/src/20200101_state_sig.csv")
out_file_df = pd.read_csv(out_file)
pd.testing.assert_frame_equal(
out_file_df,
pd.DataFrame({
"sensor_name": ["test"],
"geo_value": ["ca"],
"value": [1.5]
}))
def test_compute_ar_sensor_seed(self):
"""Test same result over 50 runs"""
values = LocationSeries(
data={
date(2020, 1, 1): -4.27815483,
date(2020, 1, 2): -4.83962077,
date(2020, 1, 3): -4.09548122,
date(2020, 1, 4): -3.86647783,
date(2020, 1, 5): -2.64494168,
date(2020, 1, 6): -3.99573135,
date(2020, 1, 7): -3.4824841,
date(2020, 1, 8): -2.77490127,
date(2020, 1, 9): -3.64162355,
date(2020, 1, 10): -2.5762891,
date(2020, 1, 11): -2.46793048,
date(2020, 1, 12): -3.20454941,
date(2020, 1, 13): -1.77057154,
date(2020, 1, 14): -0.02058535,
date(2020, 1, 15): 0.81182691,
date(2020, 1, 16): 0.32741982
})
assert len(
set(
compute_ar_sensor(date(2020, 1, 15), values, 1, 0)
for _ in range(50))) == 1
def test_compute_ar_sensor_no_regularize(self, random_normal):
"""Verified with ar.ols(x, FALSE, ar_size, intercept=TRUE, demean=FALSE)."""
random_normal.return_value = 0
values = LocationSeries(
data={
date(2020, 1, 1): -4.27815483,
date(2020, 1, 2): -4.83962077,
date(2020, 1, 3): -4.09548122,
date(2020, 1, 4): -3.86647783,
date(2020, 1, 5): -2.64494168,
date(2020, 1, 6): -3.99573135,
date(2020, 1, 7): -3.4824841,
date(2020, 1, 8): -2.77490127,
date(2020, 1, 9): -3.64162355,
date(2020, 1, 10): -2.5762891,
date(2020, 1, 11): -2.46793048,
date(2020, 1, 12): -3.20454941,
date(2020, 1, 13): -1.77057154,
date(2020, 1, 14): -0.02058535,
date(2020, 1, 15): 0.81182691,
date(2020, 1, 16): 0.32741982
})
assert np.isclose(compute_ar_sensor(date(2020, 1, 15), values, 1, 0),
-0.09105891 + 0.87530957 * -0.02058535)
assert np.isclose(
compute_ar_sensor(date(2020, 1, 15), values, 2, 0),
0.31865395 + 0.64751725 * -0.02058535 + 0.30760218 * -1.77057154)
def test_compute_ar_sensor_insufficient_data(self):
values = LocationSeries(data={
date(2020, 1, 1): -4.27815483,
date(2020, 1, 2): -4.83962077
})
assert np.isnan(compute_ar_sensor(date(2020, 1, 2), values, 1, 0))
assert np.isnan(compute_ar_sensor(date(2020, 1, 7), values, 1, 0))
def test_compute_sensors_no_covariates(self, mock_get_indicator_data,
mock_compute_ar_sensor):
"""Test only ground truth sensor is returned if no data is available to compute the rest."""
mock_get_indicator_data.return_value = {}
mock_compute_ar_sensor.return_value = 1.5
test_sensors = [
SensorConfig("a", "b", "c", 1),
SensorConfig("x", "y", "z", 2)
]
test_ground_truth_sensor = SensorConfig("i", "j", "k", 3)
test_ground_truth = [LocationSeries("ca", "state")]
assert compute_sensors(
date(2020, 5, 5), test_sensors, test_ground_truth_sensor,
test_ground_truth, False) == {
SensorConfig("i", "j", "k", 3):
[LocationSeries("ca", "state", {date(2020, 5, 2): 1.5})],
}
def test_no_results(self, mock_epidata):
mock_epidata.return_value = {"result": -2}
test_output = get_historical_sensor_data(
SensorConfig(None, None, None, None), None, None, date(2020, 1, 1),
date(2020, 1, 4))
assert test_output == (LocationSeries(None, None), [
date(2020, 1, 1),
date(2020, 1, 2),
date(2020, 1, 3),
date(2020, 1, 4)
])
def test_no_results(self, mock_epidata):
mock_epidata.return_value = [
({
"result": -2
}, {
"data_source": "src1",
"signals": "sig1",
"geo_type": "state",
"geo_value": "ca"
}),
({
"result": -2
}, {
"data_source": "src1",
"signals": "sig1",
"geo_type": "county",
"geo_value": "01001"
}),
]
test_output = get_indicator_data([
SensorConfig("src1", "sig1", None, None),
SensorConfig("src2", "sig2", None, None)
], [LocationSeries("ca", "state")], date(2020, 1, 1))
assert test_output == {}
mock_epidata.assert_called_once_with([
{
"source": "covidcast",
"data_source": "src1",
"signals": "sig1",
"time_type": "day",
"geo_type": "state",
"geo_value": "ca",
"time_values": f"{EPIDATA_START_DATE}-20200101",
"as_of": "20200101"
},
{
"source": "covidcast",
"data_source": "src2",
"signals": "sig2",
"time_type": "day",
"geo_type": "state",
"geo_value": "ca",
"time_values": f"{EPIDATA_START_DATE}-20200101",
"as_of": "20200101"
},
])
def test_error(self, mock_epidata):
mock_epidata.return_value = [({
"result": -3,
"message": "test failure"
}, {})]
with pytest.raises(Exception,
match="Bad result from Epidata: test failure"):
get_indicator_data([SensorConfig(None, None, None, None)],
[LocationSeries(None, None)], date(2020, 1, 1))
mock_epidata.assert_called_once_with([{
"source": "covidcast",
"data_source": None,
"signals": None,
"time_type": "day",
"geo_type": None,
"geo_value": None,
"time_values": f"{EPIDATA_START_DATE}-20200101",
"as_of": "20200101"
}])
def test_get_data_range_out_of_bounds(self):
test_ls = LocationSeries(data={
date(2020, 1, 1): 7,
date(2020, 1, 2): 8,
date(2020, 1, 3): 9
})
with pytest.raises(ValueError,
match="Data range must be within existing dates "
"2020-01-01 to 2020-01-03"):
test_ls.get_data_range(date(2019, 12, 31), date(2020, 1, 3))
with pytest.raises(ValueError,
match="Data range must be within existing dates "
"2020-01-01 to 2020-01-03"):
test_ls.get_data_range(date(2020, 1, 1), date(2020, 1, 4))
def test_results(self, mock_epidata):
mock_epidata.return_value = {
"result":
1,
"epidata": [{
"time_value": 20200101,
"value": 1
}, {
"time_value": 20200102,
"value": np.nan
}]
}
test_output = get_historical_sensor_data(
SensorConfig(None, None, None, None), None, None, date(2020, 1, 1),
date(2020, 1, 4))
assert test_output == (LocationSeries(None, None,
{date(2020, 1, 1): 1}), [
date(2020, 1, 2),
date(2020, 1, 3),
date(2020, 1, 4)
])
def test_historical_sensors_some_data(self, mock_historical):
"""Test non empty data is returned for first two sensors."""
mock_historical.side_effect = [
(LocationSeries(data={date(2020, 1, 1): 2}), []),
(LocationSeries(data={date(2020, 1, 3): 4}), []),
(LocationSeries(), [])
]
test_sensors = [
SensorConfig("i", "j", "k", 3),
SensorConfig("a", "b", "c", 1),
SensorConfig("x", "y", "z", 2)
]
test_ground_truth = [LocationSeries("ca", "state")]
assert historical_sensors(
None, None, test_sensors, test_ground_truth) == {
SensorConfig("i", "j", "k", 3):
[LocationSeries(data={date(2020, 1, 1): 2})],
SensorConfig("a", "b", "c", 1):
[LocationSeries(data={date(2020, 1, 3): 4})]
}
def test_compute_regression_sensor_no_data(self):
test_covariate = LocationSeries()
test_response = LocationSeries()
assert np.isnan(
compute_regression_sensor(date(2020, 1, 16), test_covariate,
test_response, False))
def test_no_data(self):
test_ls = LocationSeries()
with pytest.raises(ValueError, match="No data"):
test_ls.dates
with pytest.raises(ValueError, match="No data"):
test_ls.values
def test_compute_ar_sensor_out_of_range(self):
values = LocationSeries(data={
date(2020, 1, 1): -4.27815483,
date(2020, 1, 2): -4.83962077
})
assert np.isnan(compute_ar_sensor(date(2020, 1, 7), values, 1, 0))