New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[DOC] New estimator implementation guide #2186
Conversation
should this be called add_forecasters rather than add_estimators, or are you envisaging other sections? |
No? This is for all estimators! Forecasters are just used as the occasional example. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
under "What is my scitype?" could you first explain or link to an explanation of what a scitype is? It is not a standard term, and I do not think "Usually, the scitype of a given estimator is directly determined by what the estimator does." is sufficient. Just a sentence such as "scitype is a term we use to characterise the type of problem an estimator is designed to address" then a link to a more detailed explanation, or sth like that. Otherwise people will just google scitype and all they will find is a link to mlj.
Also, could you list the existing templates under What are sktime
extension templates?
@TonyBagnall, added clarification on this - is this better? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
looks good
* upstream/main: (79 commits) [ENH] Upgrade of `BaseTransformer` to use vectorization utility, hierarchical mtype compatibility (sktime#2219) [ENH] factory to add 3D capability to all distances exported by distances module (sktime#2051) [ENH] Relaxing `pd-multiindex` mtype to allow string instance index (sktime#2262) [ENH] refactoring `test_all_classifiers` to test class architecture (sktime#2257) [BUG] fixing `STLBootstrapTransformer` error message and docstrings (sktime#2260) [ENH] Add AutoARIMA from StatsForecast (sktime#2251) [ENH] `predict_proba` for forecasters, `tensorflow-probability` dependency (sktime#2100) [DOC] simplified extension templates for transformers and forecasters (sktime#2161) [ENH] full support for `Panel.numpyflat` mtype in `datatypes` module, checks/conversions (sktime#1772) [BUG] fix `FeatureUnion` test failure mk4 (sktime#1810) [ENH] renamed `fit-in-transform` and `fit-in-predict` to `fit_is_empty` (sktime#2250) [ENH] tag aliasing mixin for aliasing and deprecation of tags (sktime#1962) [ENH] Classifier type hints (sktime#2246) [ENH] Improved design for probabilistic error metrics (sktime#2232) remove test for StratifiedGroupKFold (sktime#2244) [DOC] typo fix and minor clarification in estimator implementation guide (sktime#2241) [DOC] improved docstrings for HIVE-COTE v1.0 (sktime#2239) [DOC] New estimator implementation guide (sktime#2186) [DOC] add scitype/mtype register pointers to docstrings in datatypes (sktime#2160) [ENH] sklearn typing util (sktime#2208) ...
This PR adds a new estimator implementation guide for developers and contributors.
The current "adding estimator" guide just displayed the forecaster extension template.
This PR contains a complete rewrite and a step-by-step guide, which should hopefully be much more helpful to new contributors.