[FEAT,BREAKING CHANGE] Add PERMBU integration to HierarchicalReconciliation class #83
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
| @@ -10,7 +10,7 @@ | |||
| [](https://pypi.org/project/hierarchicalforecast/) | |||
| [](https://github.com/Nixtla/hierarchicalforecast/blob/main/LICENSE) | |||
|
|
|||
| **HierarchicalForecast** offers a collection of reconciliation methods, including `BottomUp`, `TopDown`, `MiddleOut`, `MinTrace` and `ERM`. And Probabilistic coherent predictions including `MinTrace-normality`, `Bootstrap`, and `PERM-BU`. | |||
| **HierarchicalForecast** offers a collection of reconciliation methods, including `BottomUp`, `TopDown`, `MiddleOut`, `MinTrace` and `ERM`. And Probabilistic coherent predictions including `Normality`, `Bootstrap`, and `PERMBU`. | |||
There was a problem hiding this comment.
This change is cool, nbdev has an automated link generation if the module in quotations '``' matches something in the library.
| **Returns:**<br> | ||
| `y_tilde`: pd.DataFrame, with reconciled predictions. | ||
| """ | ||
| if intervals_method not in ['normality', 'bootstrap', 'permbu']: |
There was a problem hiding this comment.
Is the 'normality' method refering to 'mint_normality'?
I think the name could be a source of confusion.
There was a problem hiding this comment.
hey @kdgutier! I've changed the name to normality since it is a general method that works independently of the reconciliation method: https://otexts.com/fpp3/rec-prob.html
There was a problem hiding this comment.
Cool, would it be possible to explain it in the docstring with the pointer to the link?
hierarchicalforecast/methods.py
Outdated
| return _reconcile(S, P, W, y_hat, sigmah=sigmah, level=level, | ||
| bootstrap=bootstrap, bootstrap_samples=bootstrap_samples) | ||
| class Bootstrap: | ||
|
|
There was a problem hiding this comment.
It would be good to add references to Gamakumara's work for this class, in its docstring.
There was a problem hiding this comment.
Additionally, the description of the algorithm in similar fashion to that of PERMBU would be of interest to our power users.
There was a problem hiding this comment.
Lets also add some warning to PERMBU and Bootstrap on the importance of not overfitted residuals.
hierarchicalforecast/methods.py
Outdated
| temp = array.argsort(axis=1) | ||
| ranks = np.empty_like(temp) | ||
| a_range = np.arange(temp.shape[1]) | ||
| for iRow in range(temp.shape[0]): |
There was a problem hiding this comment.
I would keep caps for classes and underscores for other variables i_row?
There was a problem hiding this comment.
Sure. I can change it. To be completely transparent, I didn't change anything in the PERMBU class. I just made it compatible. The iRow variable was merged into the main branch without review in a previous PR.
hierarchicalforecast/methods.py
Outdated
| hier_levels = hier_links.shape[1]-1 | ||
| for level_idx in reversed(range(hier_levels)): | ||
| # Obtain aggregation matrix from parent/children links | ||
| children_links = np.unique(hier_links[:,level_idx:level_idx+2], |
There was a problem hiding this comment.
I wonder if there is an overlap between these children_link lines and the _get_child_nodes function.
| warnings.warn('Replacing negative forecasts with zero.') | ||
| y_hat = np.copy(y_hat) | ||
| y_hat[y_hat < 0] = 0. | ||
| # Quadratic progamming formulation |
| @@ -629,39 +753,7 @@ def __init__(self, | |||
| self.nonnegative = nonnegative | |||
| self.insample = False | |||
|
|
|||
| def reconcile(self, | |||
There was a problem hiding this comment.
I have not seen the documentation, but will showdoc(OptimalCombination.fit) work as previous?
Or will it have MinTrace
hierarchicalforecast/methods.py
Outdated
| @@ -59,23 +64,243 @@ def _reconcile(S: np.ndarray, P: np.ndarray, W: np.ndarray, | |||
| res[f'hi-{lv}'] = res['mean'] + zs * sigmah | |||
| return res | |||
|
|
|||
| # %% ../nbs/methods.ipynb 5 | |||
| class Normality: | |||
There was a problem hiding this comment.
Is this MinTraceNormality?
It would be good to add docstrings, with the shapes of sigmah is it a matrix?
There was a problem hiding this comment.
I've changed the name to normality since it is a general method that works independently of the reconciliation method: https://otexts.com/fpp3/rec-prob.html
| @@ -1,530 +0,0 @@ | |||
| { | |||
There was a problem hiding this comment.
I suggest we add the following reference:
And the following paragraph at the beginning of the notebook:
In this notebook we will explore probabilistic hierarchical sample reconciliation using the PERM-BU method. The PERM-BU (Taieb et. al, 2017) method leverages empirical bottom-level marginal distributions with empirical copula functions (describing bottom-level dependencies) to generate the distribution of the aggregate-level distributions using the bottom-up reconciliation. The sample reordering technique from the PERM-BU method reinjects multivariate dependencies into independent bottom-level samples.
sktime-disclaimer
…rarchicalforecast into feat/permbu-integration
This PR integrates the PERMBU method to the general class HierarchicalReconciliation.
The integration works for all compatible reconciliation methods (not just BottomUp).
A usage example was also included.