def test_convenience(load_df_and_hier_uv):
hd, hier = load_df_and_hier_uv
hd = hd.head(200)
ht = HTSRegressor(model='holt_winters', revision_method='OLS')
ht = ht.fit(df=hd, nodes=hier)
preds = ht.predict(steps_ahead=10)
rev = revise_forecasts(
'OLS',
nodes=ht.nodes,
forecasts=ht.hts_result.forecasts,
errors=ht.hts_result.errors,
residuals=ht.hts_result.residuals,
)
assert isinstance(rev, pandas.DataFrame)
assert numpy.allclose(preds.values, rev.values)
rev_s = revise_forecasts(
'OLS',
summing_matrix=ht.sum_mat,
forecasts=ht.hts_result.forecasts,
errors=ht.hts_result.errors,
residuals=ht.hts_result.residuals,
)
assert isinstance(rev_s, pandas.DataFrame)
assert numpy.allclose(rev.values, rev_s.values)
rev = revise_forecasts('AHP',
nodes=ht.nodes,
forecasts=ht.hts_result.forecasts)
assert isinstance(rev, pandas.DataFrame)
def test_convenience_error_handling(load_df_and_hier_uv):
hd, hier = load_df_and_hier_uv
dummy = numpy.array([1, 2, 3, 5])
with pytest.raises(ValueError):
revise_forecasts('FP', forecasts={'a': dummy}, nodes=None)
with pytest.raises(ValueError):
revise_forecasts('OLS', summing_matrix=dummy, forecasts={'a': dummy})
def test_convenience_error_handling(load_df_and_hier_uv):
_, _ = load_df_and_hier_uv
dummy = numpy.array([1, 2, 3, 5])
with pytest.raises(ValueError):
revise_forecasts("FP", forecasts={"a": dummy}, nodes=None)
with pytest.raises(ValueError):
revise_forecasts("OLS", summing_matrix=dummy, forecasts={"a": dummy})