Find and inspect the registered components.
The hub registry holds every algorithm, preprocessor, and feature component by name. These functions expose it for browsing: list by type, fetch a component's description and parameter schema, or search by keyword.
Functions
list_algorithms(type: Optional[str]=None) -> List[Dict]
List available algorithms in the registry.
Parameters
type
str
Filter by algorithm type:
- •
"classifier": Classification algorithms - •
"regressor": Regression algorithms - •
"clusterer": Clustering algorithms - •
"anomaly": Anomaly detectors - •
"associator": Association rule miners - •
None: List all algorithms
Returns
list of dict
Metadata for matching algorithms (name, description, tags).
python
>>> import tuiml
>>> classifiers = tuiml.list_algorithms(type="classifier") # doctest: +SKIP
>>> [algo["name"] for algo in classifiers[:3]] # doctest: +SKIP
['NaiveBayesClassifier', 'DecisionTreeClassifier', 'RandomForestClassifier']The names go straight into a train spec:
bash
$ tuiml.train({"model": {"name": classifiers[0]["name"]},
... "data": {"source": "iris"}}) # doctest: +SKIPdescribe_algorithm(name: str) -> Dict
Get detailed information about a specific algorithm.
Parameters
name
str
Name of the algorithm (e.g.,
"RandomForestClassifier").
Returns
dict
Metadata dictionary containing:
- •
description: Full docstring documentation - •
parameters: JSON schema for hyperparameters - •
type: Component type (classifier, etc.)
python
>>> import tuiml
>>> info = tuiml.describe_algorithm("RandomForestClassifier") # doctest: +SKIP
>>> sorted(info["parameters"]) # doctest: +SKIP
['bootstrap', 'max_depth', 'max_features', 'n_estimators', ...]search_algorithms(query: str, limit: Optional[int]=None) -> List[Dict]
Search for components by keyword in name, tags, or description.
Results are ranked by relevance, best match first. Multi-word queries are matched token-wise, so
"random forest" finds RandomForestClassifier as well as namespaced wrappers such as sklearn.RandomForestClassifier.Parameters
query
str
Search query (e.g.,
"random forest", "linear").
limit
int
= None
Maximum number of results to return.
None returns all matches.
Returns
list of dict
Metadata for matching components, best match first.
Notes
This searches every registered component type, not just algorithms - transformers and feature selectors are included in the results.
python
>>> import tuiml
>>> results = tuiml.search_algorithms("random forest", limit=3)
>>> [algo["name"] for algo in results]
['RandomForestClassifier', 'RandomForestRegressor', 'capymoa.AdaptiveRandomForest']