def revise_forecasts(
method: str,
forecasts: Dict[str, numpy.ndarray],
errors: Optional[Dict[str, numpy.ndarray]] = None,
residuals: Optional[Dict[str, numpy.ndarray]] = None,
summing_matrix: numpy.ndarray = None,
nodes: NAryTreeT = None,
transformer: TransformT = None,
):
"""
Convenience function to get revised forecast for pre-computed base forecasts
Parameters
----------
method : str
The reconciliation method to use
forecasts : Dict[str, numpy.ndarray]
A dict mapping key name to its forecasts (including in-sample forecasts). Required.
errors : Dict[str, numpy.ndarray]
A dict mapping key name to the in-sample errors. Required for methods: ``OLS``, ``WLSS``, ``WLSV``
residuals : Dict[str, numpy.ndarray]
A dict mapping key name to the residuals of in-sample forecasts. Required for methods: OLS, WLSS, WLSV
summing_matrix : numpy.ndarray
Not required if ``nodes`` argument is passed, or if using ``BU`` approach
nodes : NAryTreeT
The tree of nodes as specified in :py:class:`HierarchyTree <hts.hierarchy.HierarchyTree>`. Required if not
if using ``AHP``, ``PHA` ``FP`` methods, or if using passing the ``OLS``, ``WLSS``, ``WLSV`` methods
and not passing the ``summing_matrix`` parameter
transformer : TransformT
A transform with the method: ``inv_func`` that will be applied to the forecasts
Returns
-------
revised forecasts : pandas.DataFrame
The revised forecasts
"""
if nodes:
summing_matrix = to_sum_mat(nodes)
if method in [MethodT.AHP.name, MethodT.PHA.name, MethodT.FP.name
] and not nodes:
raise ValueError(f"Method {method} requires an NAryTree to be passed")
if method in [MethodT.OLS.name, MethodT.WLSS.name, MethodT.WLSV.name]:
if not (all([forecasts, errors, residuals]) or (not summing_matrix)):
raise ValueError(
f"Method {method} requires forecasts, errors, and residuals to be passed, as "
f"well as an NAryTree or a summing matrix")
revision = RevisionMethod(name=method,
sum_mat=summing_matrix,
transformer=transformer)
revised = revision.revise(forecasts=forecasts, mse=errors, nodes=nodes)
return pandas.DataFrame(revised, columns=list(forecasts.keys()))
def test_instantiate_revision(load_df_and_hier_visnights):
hierarchical_visnights_data, visnights_hier = load_df_and_hier_visnights
for method in ['OLS', 'FP', 'WLSS', 'WLSV', 'PHA', 'AHP', 'NONE']:
ht = HTSRegressor(model='holt_winters', revision_method=method)
ht.fit(df=hierarchical_visnights_data, nodes=visnights_hier)
rm = RevisionMethod(method,
sum_mat=ht.sum_mat,
transformer=ht.transform)
_ = ht.predict(steps_ahead=3)
revised = rm.revise(forecasts=ht.hts_result.forecasts,
mse=ht.hts_result.errors,
nodes=ht.nodes)
assert isinstance(revised, numpy.ndarray)
assert revised.shape == (11, len(ht.hts_result.forecasts))
def _init_revision(self):
self.revision_method = RevisionMethod(sum_mat=self.sum_mat,
transformer=self.transform,
name=self.method)
def revise_forecasts(
method: str,
forecasts: Dict[str, ArrayLike],
errors: Optional[Dict[str, float]] = None,
residuals: Optional[Dict[str, ArrayLike]] = None,
summing_matrix: numpy.ndarray = None,
nodes: NAryTreeT = None,
transformer: TransformT = None,
):
"""
Convenience function to get revised forecast for pre-computed base forecasts
Parameters
----------
method : str
The reconciliation method to use
forecasts : Dict[str, ArrayLike]
A dict mapping key name to its forecasts (including in-sample forecasts). Required, can be
of type ``numpy.ndarray`` of ``ndim == 1``, ``pandas.Series``, or single columned ``pandas.DataFrame``
errors : Dict[str, float]
A dict mapping key name to the in-sample errors. Required for methods: ``OLS``, ``WLSS``, ``WLSV`` if
``residuals`` is not passed
residuals : Dict[str, ArrayLike]
A dict mapping key name to the residuals of in-sample forecasts. Required for methods: ``OLS``, ``WLSS``,
``WLSV``, can be of type ``numpy.ndarray`` of ndim == 1, ``pandas.Series``, or single columned
``pandas.DataFrame``. If passing residuals, ``errors`` dict is not required and will instead be calculated
using MSE metric: ``numpy.mean(numpy.array(residual) ** 2)``
summing_matrix : numpy.ndarray
Not required if ``nodes`` argument is passed, or if using ``BU`` approach
nodes : NAryTreeT
The tree of nodes as specified in :py:class:`HierarchyTree <hts.hierarchy.HierarchyTree>`. Required if not
if using ``AHP``, ``PHA``, ``FP`` methods, or if using passing the ``OLS``, ``WLSS``, ``WLSV`` methods
and not passing the ``summing_matrix`` parameter
transformer : TransformT
A transform with the method: ``inv_func`` that will be applied to the forecasts
Returns
-------
revised forecasts : ``pandas.DataFrame``
The revised forecasts
"""
if nodes:
summing_matrix, sum_mat_labels = to_sum_mat(nodes)
if method in [MethodT.AHP.name, MethodT.PHA.name, MethodT.FP.name
] and not nodes:
raise ValueError(f"Method {method} requires an NAryTree to be passed")
if method in [MethodT.OLS.name, MethodT.WLSS.name, MethodT.WLSV.name]:
errors = _calculate_errors(method=method,
errors=errors,
residuals=residuals)
if not (all([forecasts, errors]) or (not summing_matrix)):
raise ValueError(
f"Method {method} requires forecasts, errors, and residuals to be passed, as "
f"well as an NAryTree or a summing matrix")
revision = RevisionMethod(name=method,
sum_mat=summing_matrix,
transformer=transformer)
sanitized_forecasts = _sanitize_forecasts_dict(forecasts)
revised = revision.revise(forecasts=sanitized_forecasts,
mse=errors,
nodes=nodes)
return pandas.DataFrame(revised, columns=list(sanitized_forecasts.keys()))