LassoNet

https://lasso-net.github.io/lassonet/
Admin
LassoNet is a method for feature selection in neural networks that uses a novel objective function...
Tokens:15,058
Snippets:85
Trust Score:9.7
Update:4 months ago
Show doc for...
Tokens:
Raw
### Install LassoNet Package

Source: https://lasso-net.github.io/lassonet

Use pip to install the LassoNet library. Ensure you have Python and pip installed.

```bash
pip install lassonet
```

--------------------------------

### path

Source: https://lasso-net.github.io/lassonet/api

Trains LassoNet on a lambda path, starting from lambda_start and incrementing according to path_multiplier.

```APIDOC
## path

### Description
Train LassoNet on a lambda_ path. The path is defined by the class parameters: start at lambda_start and increment according to path_multiplier. The path will stop when no feature is being used anymore. callback will be called at each step on (model, history)
If you pass lambda_seq=[], only the dense model will be trained.

### Parameters
#### Arguments
- **X** (array-like)
- **y** (array-like)
- **return_state_dicts** (bool, default=True)

### Returns
(No specific return type mentioned in the source for this method, but it implies model training and callbacks.)
```

--------------------------------

### path

Source: https://lasso-net.github.io/lassonet/api

Trains LassoNet on a lambda path, starting from lambda_start and incrementing by path_multiplier until no features are used. A callback function can be provided to be called at each step.

```APIDOC
## path

### Description
Trains LassoNet on a lambda path, starting from lambda_start and incrementing by path_multiplier until no features are used. A callback function can be provided to be called at each step.

### Parameters
* **X** - Features.
* **y** - Target variable.
* **return_state_dicts** (bool, default=True) - Whether to return state dictionaries.

### Notes
If lambda_seq=[] is passed, only the dense model will be trained.
```

--------------------------------

### LassoNetClassifier.get_params

Source: https://lasso-net.github.io/lassonet/api

Gets the parameters of the LassoNetClassifier estimator.

```APIDOC
## LassoNetClassifier.get_params

### Description
Get parameters for this estimator.

### Parameters
- **deep** (bool, default=True) - If True, will return the parameters for this estimator and contained subobjects that are estimators.

### Returns
- **params** (dict) - Parameter names mapped to their values.
```

--------------------------------

### path

Source: https://lasso-net.github.io/lassonet/api

Trains LassoNet on a lambda path. The path is defined by class parameters, starting at lambda_start and incrementing according to path_multiplier. Training stops when no feature is used. A callback function can be provided to be called at each step with the model and history.

```APIDOC
## path(_X_, _y_, _*_, _X_val=None_, _y_val=None_, _lambda_seq=None_, _lambda_max=inf_, _return_state_dicts=False_, _callback=None_, _disable_lambda_warning=False_)

### Description
Train LassoNet on a lambda path. The path is defined by the class parameters: start at lambda_start and increment according to path_multiplier. The path will stop when no feature is being used anymore. callback will be called at each step on (model, history)
If you pass lambda_seq=[], only the dense model will be trained.

### Parameters

#### Path Parameters
- **_X_** (array-like) - Training data.
- **_y_** (array-like) - Target values.
- **__** - Arbitrary arguments.
- **_X_val** (bool | None | str, optional) - Metadata routing for `X_val` parameter in `fit`. Defaults to `sklearn.utils.metadata_routing.UNCHANGED`.
- **_y_val** (bool | None | str, optional) - Metadata routing for `y_val` parameter in `fit`. Defaults to `sklearn.utils.metadata_routing.UNCHANGED`.
- **_lambda_seq** (list, optional) - If provided, only the dense model will be trained.
- **_lambda_max** (float, optional) - Maximum lambda value for the path.
- **_return_state_dicts** (bool, optional) - Whether to return state dictionaries.
- **_callback** (callable, optional) - Function to call at each step of the path training.
- **_disable_lambda_warning** (bool, optional) - Whether to disable the lambda warning.

### Returns
- **List[HistoryItem]** - A list of history items for each step in the lambda path.
```

--------------------------------

### Train LassoNet on Lambda Path

Source: https://lasso-net.github.io/lassonet/api

Train LassoNet across a sequence of lambda values to explore the regularization path. The path starts at lambda_start and increases by path_multiplier. A callback function can be provided to execute at each step.

```python
path(_X_ , _y_ , _*_ , _X_val =None_, _y_val =None_, _lambda_seq =None_, _lambda_max =inf_, _return_state_dicts =False_, _callback =None_, _disable_lambda_warning =False_)
```

--------------------------------

### path

Source: https://lasso-net.github.io/lassonet/api

Train LassoNet on a lambda path. This method computes a path of models by varying the penalty parameter lambda, starting from lambda_start and increasing according to path_multiplier.

```APIDOC
## path(X, y, *, X_val=None, y_val=None, lambda_seq=None, lambda_max=inf, return_state_dicts=False, callback=None, disable_lambda_warning=False) → List[HistoryItem]

### Description
Train LassoNet on a lambda_ path. The path is defined by the class parameters: start at lambda_start and increment according to path_multiplier. The path will stop when no feature is being used anymore. callback will be called at each step on (model, history)
If you pass lambda_seq=[], only the dense model will be trained.

### Parameters
- **X** - Features of the training data.
- **y** - Target values of the training data.
- **X_val** - Features of the validation data.
- **y_val** - Target values of the validation data.
- **lambda_seq** (iterable of float, default=None) – If specified, the model will be trained on this sequence of values.
- **lambda_max** (float, default=inf) – Maximum value for lambda.
- **return_state_dicts** (bool, default=False) – Whether to return state dictionaries.
- **callback** (callable, default=None) – A callback function called at each step on (model, history).
- **disable_lambda_warning** (bool, default=False) – Whether to disable the lambda warning.
```

--------------------------------

### Get Estimator Parameters

Source: https://lasso-net.github.io/lassonet/api

Retrieve the parameters of the LassoNet estimator. Set deep=True to include parameters of contained sub-objects.

```python
get_params(_deep =True_)
```

--------------------------------

### path

Source: https://lasso-net.github.io/lassonet/api

Trains LassoNet on a specified lambda path. The training progresses by incrementing lambda according to path_multiplier, starting from lambda_start, and stops when no more features are selected. A callback function can be provided to be executed at each training step.

```APIDOC
## path(_X_, _y_, _*_, _X_val=None_, _y_val=None_, _lambda_seq=None_, _lambda_max=inf_, _return_state_dicts=False_, _callback=None_, _disable_lambda_warning=False_)

### Description
Train LassoNet on a lambda path. The path is defined by the class parameters: start at lambda_start and increment according to path_multiplier. The path will stop when no feature is being used anymore. callback will be called at each step on (model, history)
If you pass lambda_seq=[], only the dense model will be trained.

### Parameters

#### Path Parameters
- **X** (_array-like_ of _shape_ _(__n_samples_, _n_features_)_) – Input features.
- **y** (_array-like_ of _shape_ _(__n_samples_,) or_ _(__n_samples_, _n_outputs_)_) – Target labels.
- **X_val** (_array-like_, optional) – Validation features.
- **y_val** (_array-like_, optional) – Validation labels.
- **lambda_seq** (_list_, optional) – Sequence of lambda values to train on. If empty, only the dense model is trained.
- **lambda_max** (_float_, optional) – Maximum lambda value for the path.
- **return_state_dicts** (_bool_, optional) – Whether to return state dictionaries.
- **callback** (_callable_, optional) – Function to call at each training step.
- **disable_lambda_warning** (_bool_, optional) – Whether to disable lambda warnings.
```

--------------------------------

### LassoNetCoxRegressor.get_params

Source: https://lasso-net.github.io/lassonet/api

Gets the parameters of the LassoNetCoxRegressor estimator.

```APIDOC
## get_params(_deep =True_) 

### Description
Get parameters for this estimator.

### Parameters
* **deep** (_bool_, _default=True_) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

### Returns
* **params** (_dict_) – Parameter names mapped to their values.
```

--------------------------------

### get_params

Source: https://lasso-net.github.io/lassonet/api

Get parameters for this estimator. This method returns a dictionary of parameters for the estimator and any contained sub-objects.

```APIDOC
## get_params(deep=True)

### Description
Get parameters for this estimator.

### Parameters
- **deep** (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

### Returns
- **params** (dict) – Parameter names mapped to their values.

### Return type
dict
```

--------------------------------

### get_params

Source: https://lasso-net.github.io/lassonet/api

Gets the parameters of the estimator, including those of nested objects.

```APIDOC
## get_params

### Description
Gets the parameters of the estimator, including those of nested objects.

### Parameters
* **deep** (bool, default=True) - If True, will return the parameters for this estimator and contained subobjects that are estimators.

### Returns
* **params** (dict) - Parameter names mapped to their values.
```

--------------------------------

### LassoNetRegressorCV.path

Source: https://lasso-net.github.io/lassonet/api

Trains LassoNet on a lambda path, starting from lambda_start and incrementing according to path_multiplier until no features are used.

```APIDOC
## LassoNetRegressorCV.path

### Description
Trains LassoNet on a lambda path, starting from lambda_start and incrementing according to path_multiplier until no features are used. A callback can be provided to be called at each step with the model and history.

### Method
path(X, y, *, return_state_dicts=True)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Request Example
None

### Response
#### Success Response (200)
- **oracle** (int) – Ideal number of features to select.
- **order** (LongTensor) – Order of the selected features.
- **wrong** (Tensor) – Expected number of wrong features.
- **paths** (List[List[HistoryItem]])
- **prob** (torch.Tensor) – Tensor of shape (n_steps, n_features) containing the selection probability of each feature at lambda value.

#### Response Example
None
```

--------------------------------

### get_params

Source: https://lasso-net.github.io/lassonet/api

Gets the parameters for this estimator, including those of nested objects if `deep` is True.

```APIDOC
## get_params

### Description
Get parameters for this estimator.

### Parameters
#### Arguments
- **deep** (bool, default=True) - If True, will return the parameters for this estimator and contained subobjects that are estimators.

### Returns
- **params** (dict) - Parameter names mapped to their values.

### Return type
dict
```

--------------------------------

### LassoNetRegressor Initialization

Source: https://lasso-net.github.io/lassonet/api

Initialize the LassoNetRegressor with various parameters to control network architecture, regularization, training, and more. Default values are provided for most parameters.

```python
lassonet.LassoNetRegressor(_*_ , _hidden_dims =(100,)_, _lambda_start ='auto'_, _lambda_seq =None_, _gamma =0.0_, _gamma_skip =0.0_, _path_multiplier =1.02_, _M =10_, _groups =None_, _dropout =0_, _batch_size =None_, _optim =None_, _n_iters =(1000, 100)_, _patience =(100, 10)_, _tol =0.99_, _backtrack =False_, _val_size =None_, _device =None_, _verbose =1_, _random_state =None_, _torch_seed =None_)
```

--------------------------------

### LassoNetCoxRegressorCV

Source: https://lasso-net.github.io/lassonet/api

Initializes the LassoNetCoxRegressorCV model. It inherits parameters from BaseLassoNet and includes a cross-validation generator.

```APIDOC
## LassoNetCoxRegressorCV

### Description
Initializes the LassoNetCoxRegressorCV model. It inherits parameters from BaseLassoNet and includes a cross-validation generator.

### Parameters
* **cv** (int, cross-validation generator or iterable, default=None) - Determines the cross-validation splitting strategy. Default is 5-fold cross-validation. See <https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.check_cv.html>
* **kwargs** - Additional keyword arguments passed to the base class.
```

--------------------------------

### lassonet.plot_path

Source: https://lasso-net.github.io/lassonet/api

Plots the evolution of the model on the path, including lambda, number of selected variables, and score. Requires that model.path(return_state_dicts=True) has been called beforehand.

```APIDOC
## lassonet.plot_path

### Description
Plot the evolution of the model on the path, namely: - lambda - number of selected variables - score
Requires to have called model.path(return_state_dicts=True) beforehand.

### Parameters
* **model** (LassoNetClassifier or LassoNetRegressor)
* **X_test** (array-like)
* **y_test** (array-like)
* **score_function** (function or None) – if None, use score_function=model.score score_function must take as input X_test, y_test
```

--------------------------------

### stability_selection

Source: https://lasso-net.github.io/lassonet/api

Computes stability selection paths to be passed to lassonet.utils.selection_probability.

```APIDOC
## stability_selection

### Description
Compute stability selection paths to be passed to lassonet.utils.selection_probability.

### Parameters
#### Arguments
- **X** (array-like of shape (n_samples, n_features)) - Training data
- **y** (array-like of shape (n_samples,) or (n_samples, n_outputs)) - Target values
- **n_models** (int) - Number of models to train

### Returns
- **oracle** (int) - Ideal number of features to select.
- **order** (LongTensor) - Order of the selected features.
- **wrong** (Tensor) - Expected number of wrong features.
- **paths** (List[List[HistoryItem]])
- **prob** (torch.Tensor) - Tensor of shape (n_steps, n_features) containing the selection probability of each feature at lambda value.

### Return type
Tuple[List[List[HistoryItem]], Tensor, Tuple[Tensor, LongTensor]]
```

--------------------------------

### LassoNetClassifier.fit

Source: https://lasso-net.github.io/lassonet/api

Trains the LassoNet classifier model.

```APIDOC
## LassoNetClassifier.fit

### Description
Train the model. Note that if lambda_ is not given, the trained model will most likely not use any feature. If dense_only is True, will only train a dense model.

### Parameters
- **X** (array-like) - Training data.
- **y** (array-like) - Target values.
- **X_val** (array-like, optional) - Validation data.
- **y_val** (array-like, optional) - Validation targets.
- **dense_only** (bool, default=False) - If True, only train a dense model.
```

--------------------------------

### set_fit_request

Source: https://lasso-net.github.io/lassonet/api

Configures metadata routing for the `fit` method. This is relevant when using LassoNet within a meta-estimator, allowing control over which metadata is passed to the `fit` method.

```APIDOC
## set_fit_request(_*_, _X_val=sklearn.utils.metadata_routing.UNCHANGED_, _dense_only=sklearn.utils.metadata_routing.UNCHANGED_, _y_val=sklearn.utils.metadata_routing.UNCHANGED_)

### Description
Request metadata passed to the `fit` method. Note that this method is only relevant if `enable_metadata_routing=True` (see `sklearn.set_config()`). Please see User Guide on how the routing mechanism works.

### Parameters

#### Path Parameters
- **X_val** (_str_, _True_, _False_, or _None_, optional) – Metadata routing for `X_val` parameter in `fit`.
- **dense_only** (_str_, _True_, _False_, or _None_, optional) – Metadata routing for `dense_only` parameter in `fit`.
- **y_val** (_str_, _True_, _False_, or _None_, optional) – Metadata routing for `y_val` parameter in `fit`.

### Returns
- **self** (_object_) – The updated object.
```

--------------------------------

### fit

Source: https://lasso-net.github.io/lassonet/api

Train the model. This method trains the LassoNet regressor on the provided data. If lambda_ is not specified, the trained model may not utilize any features.

```APIDOC
## fit(X, y, *, X_val=None, y_val=None, dense_only=False)

### Description
Train the model. Note that if lambda_ is not given, the trained model will most likely not use any feature. If dense_only is True, will only train a dense model.

### Parameters
- **X** - Features of the training data.
- **y** - Target values of the training data.
- **X_val** - Features of the validation data.
- **y_val** - Target values of the validation data.
- **dense_only** (bool, default=False) - If True, only train a dense model.
```

--------------------------------

### Train LassoNet Model

Source: https://lasso-net.github.io/lassonet/api

Train the LassoNet model using the fit method. Feature selection is performed based on the lambda parameter. Setting dense_only to True trains only a dense model.

```python
fit(_X_ , _y_ , _*_ , _X_val =None_, _y_val =None_, _dense_only =False_)
```

--------------------------------

### stability_selection

Source: https://lasso-net.github.io/lassonet/api

Computes stability selection paths, which are intended to be passed to `lassonet.utils.selection_probability`. This method trains multiple models to generate these paths.

```APIDOC
## stability_selection(_X_, _y_, _n_models=20_)

### Description
Compute stability selection paths to be passed to lassonet.utils.selection_probability.

### Parameters

#### Path Parameters
- **_X** (array-like of shape (n_samples, n_features)) – Training data.
- **_y** (array-like of shape (n_samples,) or (n_samples, n_outputs)) – Target values.
- **_n_models** (int, optional) – Number of models to train. Defaults to 20.

### Returns
- **Tuple[List[List[HistoryItem]], Tensor, Tuple[Tensor, LongTensor]]** - A tuple containing:
    - A list of lists of HistoryItem objects for each model.
    - A Tensor.
    - A tuple containing a Tensor and a LongTensor.
```

--------------------------------

### LassoNetCoxRegressor.get_metadata_routing

Source: https://lasso-net.github.io/lassonet/api

Retrieves the metadata routing configuration for the LassoNetCoxRegressor object.

```APIDOC
## get_metadata_routing() 

### Description
Get metadata routing of this object.
Please check User Guide on how the routing mechanism works.

### Returns
* **routing** (_MetadataRequest_) – A `MetadataRequest` encapsulating routing information.
```

--------------------------------

### LassoNetClassifier.get_metadata_routing

Source: https://lasso-net.github.io/lassonet/api

Retrieves the metadata routing configuration for the LassoNetClassifier.

```APIDOC
## LassoNetClassifier.get_metadata_routing

### Description
Get metadata routing of this object. Please check User Guide on how the routing mechanism works.

### Returns
- **routing** (MetadataRequest) - A `MetadataRequest` encapsulating routing information.
```

--------------------------------

### fit

Source: https://lasso-net.github.io/lassonet/api

Trains the model. If lambda_ is not provided, the trained model may not use any features.

```APIDOC
## fit

### Description
Train the model. Note that if lambda_ is not given, the trained model will most likely not use any feature.

### Parameters
#### Arguments
- **X** (array-like)
- **y** (array-like)
```

--------------------------------

### stability_selection

Source: https://lasso-net.github.io/lassonet/api

Computes stability selection paths for use with lassonet.utils.selection_probability. It helps in determining the ideal number of features to select and their order.

```APIDOC
## stability_selection

### Description
Compute stability selection paths to be passed to lassonet.utils.selection_probability.

### Parameters
- **X** (array-like of shape (n_samples, n_features)) - Training data
- **y** (array-like of shape (n_samples,) or (n_samples, n_outputs)) - Target values
- **n_models** (int) - Number of models to train

### Returns
- **oracle** (int) - Ideal number of features to select.
- **order** (LongTensor) - Order of the selected features.
- **wrong** (Tensor) - Expected number of wrong features.
- **paths** (List[List[HistoryItem]])
- **prob** (torch.Tensor) - Tensor of shape (n_steps, n_features) containing the selection probability of each feature at lambda value.
```

--------------------------------

### stability_selection

Source: https://lasso-net.github.io/lassonet/api

Computes stability selection paths, which are used for passing to `lassonet.utils.selection_probability`.

```APIDOC
## stability_selection(_X_, _y_, _n_models =20_) -> Tuple[List[List[HistoryItem]], Tensor, Tuple[Tensor, LongTensor]]

### Description
Compute stability selection paths to be passed to lassonet.utils.selection_probability.

### Parameters
#### Path Parameters
* **X** (_array-like_ of shape _(__n_samples_, _n_features_)_) – Training data
* **y** (_array-like_ of shape _(__n_samples_, _) or _(__n_samples_, _n_outputs_)_) – Target values
* **n_models** (_int_) – Number of models to train
```

--------------------------------

### LassoNet Citation

Source: https://lasso-net.github.io/lassonet

Cite this paper when using LassoNet in your research. This is the primary reference for the algorithms and methods.

```bibtex
@article{lemhadri2019neural,
  title={LassoNet: Neural Networks with Feature Sparsity},
  author={Lemhadri, Ismael and Ruan, Feng and
                Abraham, Louis and Tibshirani, Robert},
  journal={arXiv preprint arXiv:1907.12207},
  year={2019}
}

```

--------------------------------

### LassoNetRegressorCV.get_params

Source: https://lasso-net.github.io/lassonet/api

Retrieves the parameters of the estimator, including those of its contained sub-objects.

```APIDOC
## LassoNetRegressorCV.get_params

### Description
Retrieves the parameters of the estimator, including those of its contained sub-objects.

### Method
get_params(deep=True)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Request Example
None

### Response
#### Success Response (200)
- **params** (dict) - Parameter names mapped to their values.

#### Response Example
None
```

--------------------------------

### fit

Source: https://lasso-net.github.io/lassonet/api

Trains the LassoNetCoxRegressorCV model using the provided data. If lambda_ is not specified, the model may not select any features.

```APIDOC
## fit

### Description
Trains the LassoNetCoxRegressorCV model using the provided data. If lambda_ is not specified, the model may not select any features.

### Parameters
* **X** - Features.
* **y** - Target variable.
```

--------------------------------

### set_params

Source: https://lasso-net.github.io/lassonet/api

Sets the parameters of the estimator. This method works for simple estimators as well as nested objects like Pipelines, allowing parameter updates for components using the '<component>__<parameter>' syntax.

```APIDOC
## set_params(**params)

### Description
Set the parameters of this estimator.
The method works on simple estimators as well as on nested objects (such as `Pipeline`). The latter have parameters of the form `<component>__<parameter>` so that it’s possible to update each component of a nested object.

### Parameters

#### Path Parameters
- ****params** (dict) – Estimator parameters.

### Returns
- **self** – Estimator instance.

### Return type
- estimator instance
```

--------------------------------

### LassoNetRegressorCV.fit

Source: https://lasso-net.github.io/lassonet/api

Trains the LassoNet model. If lambda_ is not provided, the model may not use any features.

```APIDOC
## LassoNetRegressorCV.fit

### Description
Trains the LassoNet model. If lambda_ is not provided, the model may not use any features.

### Method
fit(X, y)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Request Example
None

### Response
#### Success Response (200)
None

#### Response Example
None
```

--------------------------------

### lassonet.lassonet_path

Source: https://lasso-net.github.io/lassonet/api

Computes the LassoNet path for classification or regression tasks. It can optionally use validation data to monitor performance during path computation.

```APIDOC
## lassonet.lassonet_path

### Parameters
* **X** (array-like of shape (n_samples, n_features)) – Training data
* **y** (array-like of shape (n_samples,) or (n_samples, n_outputs)) – Target values
* **task** (str, must be "classification" or "regression") – Task
* **X_val** (array-like of shape (n_samples, n_features)) – Validation data
* **y_val** (array-like of shape (n_samples,) or (n_samples, n_outputs)) – Validation values
* **parameters.** (See BaseLassoNet for the other)
```

--------------------------------

### set_fit_request

Source: https://lasso-net.github.io/lassonet/api

Requests metadata to be passed to the `fit` method. This is relevant when `enable_metadata_routing=True`. It allows specifying how metadata for parameters like `X_val` and `dense_only` should be handled.

```APIDOC
## set_fit_request(_*_, _X_val=sklearn.utils.metadata_routing.UNCHANGED_, _dense_only=sklearn.utils.metadata_routing.UNCHANGED_, _y_val=sklearn.utils.metadata_routing.UNCHANGED_)

### Description
Request metadata passed to the `fit` method. Note that this method is only relevant if `enable_metadata_routing=True` (see `sklearn.set_config()`). Please see User Guide on how the routing mechanism works.

The options for each parameter are:
  * `True`: metadata is requested, and passed to `fit` if provided. The request is ignored if metadata is not provided.
  * `False`: metadata is not requested and the meta-estimator will not pass it to `fit`.
  * `None`: metadata is not requested, and the meta-estimator will raise an error if the user provides it.
  * `str`: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (`sklearn.utils.metadata_routing.UNCHANGED`) retains the existing request. This allows you to change the request for some parameters and not others.
Added in version 1.3.
Note
This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a `Pipeline`. Otherwise it has no effect.

### Parameters

#### Path Parameters
- **_X_val** (str | True | False | None, optional) – Metadata routing for `X_val` parameter in `fit`. Defaults to `sklearn.utils.metadata_routing.UNCHANGED`.
- **_dense_only** (str | True | False | None, optional) – Metadata routing for `dense_only` parameter in `fit`. Defaults to `sklearn.utils.metadata_routing.UNCHANGED`.
- **_y_val** (str | True | False | None, optional) – Metadata routing for `y_val` parameter in `fit`. Defaults to `sklearn.utils.metadata_routing.UNCHANGED`.

### Returns
- **self** – The updated object.

### Return type
- object
```

--------------------------------

### LassoNetCoxRegressor.fit

Source: https://lasso-net.github.io/lassonet/api

Trains the LassoNet Cox regression model using the provided training data.

```APIDOC
## fit(_X_ , _y_ , _*_ , _X_val =None_, _y_val =None_, _dense_only =False_) 

### Description
Train the model. Note that if lambda_ is not given, the trained model will most likely not use any feature. If dense_only is True, will only train a dense model. 

### Parameters
* **X** (_array-like_) – Training data.
* **y** (_array-like_) – Target values.
* **X_val** (_array-like, optional_) – Validation data for early stopping.
* **y_val** (_array-like, optional_) – Validation target values for early stopping.
* **dense_only** (_bool, optional_) – If True, only train a dense model.
```

--------------------------------

### stability_selection

Source: https://lasso-net.github.io/lassonet/api

Computes stability selection paths for LassoNet. This function is used to determine the optimal number of features to select by training multiple models and analyzing their stability.

```APIDOC
## stability_selection(_X_ , _y_ , _n_models =20_) 

### Description
Compute stability selection paths to be passed to lassonet.utils.selection_probability.

### Parameters
* **X** (_array-like_ _of_ _shape_ _(__n_samples_ _,__n_features_ _)_) – Training data
* **y** (_array-like_ _of_ _shape_ _(__n_samples_ _,__) or_ _(__n_samples_ _,__n_outputs_ _)_) – Target values
* **n_models** (_int_) – Number of models to train

### Returns
* **paths** (_List[List[HistoryItem]]_)
* **prob** (_torch.Tensor_) – Tensor of shape (n_steps, n_features) containing the selection probability of each feature at lambda value.
* **oracle** (_int_) – Ideal number of features to select.
* **order** (_LongTensor_) – Order of the selected features.
* **wrong** (_Tensor_) – Expected number of wrong features.
```

--------------------------------

### set_params

Source: https://lasso-net.github.io/lassonet/api

Sets the parameters of the estimator. This method works for simple estimators as well as nested objects.

```APIDOC
## set_params(**params)

### Description
Set the parameters of this estimator.
The method works on simple estimators as well as on nested objects (such as `Pipeline`). The latter have parameters of the form `<component>__<parameter>` so that it’s possible to update each component of a nested object.

### Parameters
#### Path Parameters
* ****params** (_dict_) – Estimator parameters.

### Returns
**self** (_estimator instance_) – Estimator instance.
```

--------------------------------

### set_params

Source: https://lasso-net.github.io/lassonet/api

Sets the parameters of the estimator, applicable to simple estimators and nested objects.

```APIDOC
## set_params(_** params_)

### Description
Set the parameters of this estimator. The method works on simple estimators as well as on nested objects (such as `Pipeline`). The latter have parameters of the form `<component>__<parameter>` so that it’s possible to update each component of a nested object.

### Parameters
#### Path Parameters
- ****params** (_dict_) – Estimator parameters.

### Returns
#### Success Response (200)
- **self** (_estimator instance_) – Estimator instance.
```

--------------------------------

### set_score_request

Source: https://lasso-net.github.io/lassonet/api

Configures metadata routing for the `score` method. This is useful when LassoNet is part of a meta-estimator, enabling control over the passing of metadata to the `score` method.

```APIDOC
## set_score_request(_*_, _sample_weight=sklearn.utils.metadata_routing.UNCHANGED_)

### Description
Request metadata passed to the `score` method. Note that this method is only relevant if `enable_metadata_routing=True` (see `sklearn.set_config()`). Please see User Guide on how the routing mechanism works.

### Parameters

#### Path Parameters
- **sample_weight** (_str_, _True_, _False_, or _None_, optional) – Metadata routing for `sample_weight` parameter in `score`.

### Returns
- **self** (_object_) – The updated object.
```

--------------------------------

### get_metadata_routing

Source: https://lasso-net.github.io/lassonet/api

Retrieves the metadata routing configuration for the estimator.

```APIDOC
## get_metadata_routing

### Description
Retrieves the metadata routing configuration for the estimator. Please check User Guide on how the routing mechanism works.

### Returns
* **routing** (MetadataRequest) - A `MetadataRequest` encapsulating routing information.
```

--------------------------------

### set_fit_request

Source: https://lasso-net.github.io/lassonet/api

Requests metadata to be passed to the `fit` method. This is relevant when `enable_metadata_routing=True` and the estimator is used within a meta-estimator like a Pipeline.

```APIDOC
## set_fit_request(_*, _X_val: bool | None | str = '$UNCHANGED$', _dense_only: bool | None | str = '$UNCHANGED$', _y_val: bool | None | str = '$UNCHANGED$') -> LassoNetClassifierCV

### Description
Request metadata passed to the `fit` method. Note that this method is only relevant if `enable_metadata_routing=True` (see `sklearn.set_config()`). Please see User Guide on how the routing mechanism works.

The options for each parameter are:
  * `True`: metadata is requested, and passed to `fit` if provided. The request is ignored if metadata is not provided.
  * `False`: metadata is not requested and the meta-estimator will not pass it to `fit`.
  * `None`: metadata is not requested, and the meta-estimator will raise an error if the user provides it.
  * `str`: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (`sklearn.utils.metadata_routing.UNCHANGED`) retains the existing request. This allows you to change the request for some parameters and not others.
Added in version 1.3.
Note
This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a `Pipeline`. Otherwise it has no effect.

### Parameters
#### Path Parameters
* **X_val** (_str_, _True_, _False_, or _None_, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for `X_val` parameter in `fit`.
* **dense_only** (_str_, _True_, _False_, or _None_, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for `dense_only` parameter in `fit`.
* **y_val** (_str_, _True_, _False_, or _None_, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for `y_val` parameter in `fit`.

### Returns
**self** (_object_) – The updated object.
```

--------------------------------

### LassoNetClassifier

Source: https://lasso-net.github.io/lassonet/api

Implements the LassoNet algorithm as a classifier. It allows for regularization and feature selection within a neural network structure.

```APIDOC
## LassoNetClassifier

### Description
Use LassoNet as classifier.

### Parameters
- **class_weight** (iterable of float, default=None) - Weights for different classes in training.
- **hidden_dims** (tuple of int, default=(100,)) - Shape of the hidden layers.
- **lambda_start** (float, default='auto') - First value on the path. Leave ‘auto’ to estimate it automatically.
- **lambda_seq** (iterable of float) - Sequence of lambda values for training.
- **gamma** (float, default=0.0) - l2 penalization on the network.
- **gamma_skip** (float, default=0.0) - l2 penalization on the skip connection.
- **path_multiplier** (float, default=1.02) - Multiplicative factor to increase the penalty parameter over the path.
- **M** (float, default=10.0) - Hierarchy parameter.
- **groups** (None or list of lists, default=None) - Use group LassoNet regularization.
- **dropout** (float, default=None)
- **batch_size** (int, default=None) - Batch size for training.
- **optim** (torch optimizer or tuple of 2 optimizers, default=None) - Optimizer for initial training and path computation.
- **n_iters** (int or pair of int, default=(1000, 100)) - Maximum number of training epochs.
- **patience** (int or pair of int or None, default=(100, 10)) - Number of epochs to wait without improvement during early stopping.
- **tol** (float, default=0.99) - Minimum improvement for early stopping.
- **backtrack** (bool, default=False) - Ensures the objective function decreases.
- **val_size** (float, default=None) - Proportion of data to use for early stopping.
- **device** (torch device, default=None) - Device for training.
- **verbose** (int, default=1)
- **random_state** - Random state for validation
- **torch_seed** - Torch state for model random initialization
```

--------------------------------

### set_fit_request

Source: https://lasso-net.github.io/lassonet/api

Requests metadata to be passed to the fit method, relevant when enable_metadata_routing is True.

```APIDOC
## set_fit_request(_*_ , _X_val : bool | None | str = '$UNCHANGED$'_, _dense_only : bool | None | str = '$UNCHANGED$'_, _y_val : bool | None | str = '$UNCHANGED$'_) → LassoNetRegressor

### Description
Request metadata passed to the `fit` method. Note that this method is only relevant if `enable_metadata_routing=True` (see `sklearn.set_config()`). Please see User Guide on how the routing mechanism works. The options for each parameter are: `True`: metadata is requested, and passed to `fit` if provided. The request is ignored if metadata is not provided. `False`: metadata is not requested and the meta-estimator will not pass it to `fit`. `None`: metadata is not requested, and the meta-estimator will raise an error if the user provides it. `str`: metadata should be passed to the meta-estimator with this given alias instead of the original name. The default (`sklearn.utils.metadata_routing.UNCHANGED`) retains the existing request. This allows you to change the request for some parameters and not others. Added in version 1.3. Note This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a `Pipeline`. Otherwise it has no effect.

### Parameters
#### Path Parameters
- **X_val** (_str_ _,__True_ _,__False_ _, or_ _None_ _,__default=sklearn.utils.metadata_routing.UNCHANGED_) – Metadata routing for `X_val` parameter in `fit`.
- **dense_only** (_str_ _,__True_ _,__False_ _, or_ _None_ _,__default=sklearn.utils.metadata_routing.UNCHANGED_) – Metadata routing for `dense_only` parameter in `fit`.
- **y_val** (_str_ _,__True_ _,__False_ _, or_ _None_ _,__default=sklearn.utils.metadata_routing.UNCHANGED_) – Metadata routing for `y_val` parameter in `fit`.

### Returns
#### Success Response (200)
- **self** (_object_) – The updated object.
```

--------------------------------

### set_params

Source: https://lasso-net.github.io/lassonet/api

Sets the parameters of the estimator, including those of nested objects.

```APIDOC
## set_params

### Description
Sets the parameters of the estimator, including those of nested objects.

### Parameters
* **params** (dict) - Estimator parameters.

### Returns
* **self** (estimator instance) - Estimator instance.
```

--------------------------------

### LassoNetRegressorCV.set_fit_request

Source: https://lasso-net.github.io/lassonet/api

Requests metadata to be passed to the `fit` method. This is relevant when using the estimator within a meta-estimator and metadata routing is enabled.

```APIDOC
## LassoNetRegressorCV.set_fit_request

### Description
Requests metadata to be passed to the `fit` method. This is relevant when using the estimator within a meta-estimator and metadata routing is enabled.

### Method
set_fit_request(*, X_val=None, dense_only=None, y_val=None)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Request Example
None

### Response
#### Success Response (200)
- **LassoNetRegressorCV** (LassoNetRegressorCV) - The estimator instance with updated fit request.

#### Response Example
None
```

--------------------------------

### LassoNetRegressorCV.get_metadata_routing

Source: https://lasso-net.github.io/lassonet/api

Retrieves the metadata routing for this object, which governs how metadata is passed to nested estimators.

```APIDOC
## LassoNetRegressorCV.get_metadata_routing

### Description
Retrieves the metadata routing for this object, which governs how metadata is passed to nested estimators.

### Method
get_metadata_routing()

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Request Example
None

### Response
#### Success Response (200)
- **routing** (MetadataRequest) - A `MetadataRequest` encapsulating routing information.

#### Response Example
None
```

--------------------------------

### set_fit_request

Source: https://lasso-net.github.io/lassonet/api

Requests metadata for the `fit` method, allowing configuration of which metadata is passed.

```APIDOC
## set_fit_request

### Description
Requests metadata for the `fit` method, allowing configuration of which metadata is passed. This method is only relevant if `enable_metadata_routing=True`.

### Parameters
* **X_val** (str | bool | None, default=sklearn.utils.metadata_routing.UNCHANGED) - Metadata routing for `X_val` parameter in `fit`.
* **dense_only** (str | bool | None, default=sklearn.utils.metadata_routing.UNCHANGED) - Metadata routing for `dense_only` parameter in `fit`.
* **y_val** (str | bool | None, default=sklearn.utils.metadata_routing.UNCHANGED) - Metadata routing for `y_val` parameter in `fit`.

### Returns
* **self** (object) - The updated object.
```

--------------------------------

### stability_selection

Source: https://lasso-net.github.io/lassonet/api

Computes stability selection paths to be passed to lassonet.utils.selection_probability. This function is useful for understanding feature selection stability across different model iterations.

```APIDOC
## stability_selection

### Description
Compute stability selection paths to be passed to lassonet.utils.selection_probability.

### Parameters
* **X** (array-like of shape (n_samples, n_features)) – Training data
* **y** (array-like of shape (n_samples,) or (n_samples, n_outputs)) – Target values
* **n_models** (int) – Number of models to train

### Returns
* **oracle** (int) – Ideal number of features to select.
* **order** (LongTensor) – Order of the selected features.
* **wrong** (Tensor) – Expected number of wrong features.
* **paths** (List[List[HistoryItem]])
* **prob** (torch.Tensor) – Tensor of shape (n_steps, n_features) containing the selection probability of each feature at lambda value.
```

--------------------------------

### set_params

Source: https://lasso-net.github.io/lassonet/api

Sets the parameters of the estimator. This method can be used on simple estimators as well as nested objects like Pipelines.

```APIDOC
## set_params

### Description
Sets the parameters of this estimator.
The method works on simple estimators as well as on nested objects (such as `Pipeline`). The latter have parameters of the form `<component>__<parameter>` so that it’s possible to update each component of a nested object.

### Parameters
#### Keyword Arguments
- **params** (dict) - Estimator parameters.

### Returns
- **self** (estimator instance) - Estimator instance.

### Return type
estimator instance
```

--------------------------------

### set_params

Source: https://lasso-net.github.io/lassonet/api

Sets the parameters of the estimator. This method can be used for both simple estimators and nested objects like Pipelines, allowing for parameter updates at different levels.

```APIDOC
## set_params(**params)

### Description
Set the parameters of this estimator. The method works on simple estimators as well as on nested objects (such as `Pipeline`). The latter have parameters of the form `<component>__<parameter>` so that it’s possible to update each component of a nested object.

### Parameters

#### Path Parameters
- **params** (_dict_) – Estimator parameters.

### Returns
- **self** (_estimator instance_) – Estimator instance.
```
Tokens:
Raw