def test_predict(input_df):
model = HourlyDayOfWeekModel(min_contiguous_months=0)
model.fit(input_df)
# Test cases on the output data types of predict function.
# When summed = True, prediction and variance are atomic
# floating point numbers
prediction, variance= model.predict(input_df, summed=True)
assert type(prediction) == np.float64
#assert type(variance) == np.float64
# When summed=False, prediction and variance are pandas Series.
prediction, variance= model.predict(input_df, summed=False)
assert type(prediction) == pd.Series
assert type(variance) == pd.Series
# We expect out linear regression model to make prediction between 0.95 & 2.0 because the energy
# consumed in input_df dataframe is always 1.0, please take a look at hourly_trace_with_dummy_energy function
assert prediction[0].item() > 0.95 and prediction[0].item() < 2.0
def test_min_contiguous_months(input_df):
min_contiguous_months = 9
model = HourlyDayOfWeekModel(min_contiguous_months=min_contiguous_months)
with pytest.raises(model_exceptions.DataSufficiencyException) as sufficiency_exception:
model.fit(input_df)