pylwl.models package¶

pylwl.models.base_model module¶

class pylwl.models.base_model.BaseModel[source]¶

Bases: BaseEstimator

Base class for all models

evaluate(y_true, y_pred, list_metrics=None)[source]¶

Evaluate the model using specified metrics.

Parameters:

y_true (array-like) – True target values.
y_pred (array-like) – Model’s predicted values.
list_metrics (list of str, optional) – Names of metrics for evaluation (e.g., “MSE”, “MAE”).

Returns:

Evaluation metrics and their values.

Return type:

dict

fit(X, y)[source]¶: Fit the model to the training data.

static load_model(load_path='history', filename='model.pkl')[source]¶

Load a model from a pickle file.

Parameters:

load_path (str, optional) – Path to load the model from (default: “history”).
filename (str, optional) – Filename of the saved model (default: “model.pkl”).

Returns:

The loaded model.

Return type:

BaseMlp

predict(X)[source]¶: Predict using the trained model.

save_evaluation_metrics(y_true, y_pred, list_metrics=('RMSE', 'MAE'), save_path='history', filename='metrics.csv')[source]¶

Save evaluation metrics to a CSV file.

Parameters:

y_true (array-like) – Ground truth values.
y_pred (array-like) – Model predictions.
list_metrics (list of str, optional) – Metrics for evaluation (default: (“RMSE”, “MAE”)).
save_path (str, optional) – Path to save the file (default: “history”).
filename (str, optional) – Filename for saving metrics (default: “metrics.csv”).

save_model(save_path='history', filename='model.pkl')[source]¶

Save the trained model to a pickle file.

Parameters:

save_path (str, optional) – Path to save the model (default: “history”).
filename (str, optional) – Filename for saving model, with “.pkl” extension (default: “model.pkl”).

save_training_loss(save_path='history', filename='loss.csv')[source]¶

Save training loss history to a CSV file.

Parameters:

save_path (str, optional) – Path to save the file (default: “history”).
filename (str, optional) – Filename for saving loss history (default: “loss.csv”).

save_y_predicted(X, y_true, save_path='history', filename='y_predicted.csv')[source]¶

Save true and predicted values to a CSV file.

Parameters:

X (array-like or torch.Tensor) – Input features.
y_true (array-like) – True values.
save_path (str, optional) – Path to save the file (default: “history”).
filename (str, optional) – Filename for saving predicted values (default: “y_predicted.csv”).

score(X, y)[source]¶: Evaluate the model’s performance on the given data.

pylwl.models.classic_lw module¶

class pylwl.models.classic_lw.BaseLW(kernel='gaussian', tau=1.0)[source]¶

Bases: BaseModel

Base class for locally weighted models.

This class provides the foundation for locally weighted regression and classification models. It includes methods for computing kernel weights based on a specified kernel function.

Parameters:

kernel (str or callable, optional) – The kernel function to use. If a string is provided, it should match the name of a kernel function in the kernel_module. If a callable is provided, it should accept distances and tau as arguments and return weights.
tau (float, optional) – The bandwidth parameter for the kernel function (default: 1.0).

kernel¶

The kernel function used for computing weights.

Type:: str or callable

kernel_func_¶

The resolved kernel function (either from kernel_module or the provided callable).

Type:: callable

tau¶

The bandwidth parameter for the kernel function.

Type:: float

class pylwl.models.classic_lw.LwClassifier(kernel='gaussian', tau=1.0)[source]¶

Bases: BaseLW, ClassifierMixin

Locally Weighted Classifier.

This class implements a locally weighted classification model using a specified kernel function and bandwidth parameter. It predicts class probabilities and labels by fitting a weighted linear model for each query point.

Parameters:

kernel (str or callable, optional) – The kernel function to use. If a string is provided, it should match the name of a kernel function in the kernel_module. If a callable is provided, it should accept distances and tau as arguments and return weights.
tau (float, optional) – The bandwidth parameter for the kernel function (default: 1.0).

X_¶

The training data.

Type:: ndarray, shape (n_samples, n_features)

y_raw_¶

The raw target values for the training data.

Type:: ndarray, shape (n_samples,)

classes_¶

The unique class labels.

Type:: ndarray, shape (n_classes,)

n_classes_¶

The number of unique classes.

Type:: int

lb_¶

The label binarizer used for encoding class labels.

Type:: LabelBinarizer

y_bin_¶

The binarized target values for the training data.

Type:: ndarray, shape (n_samples, n_classes) or (n_samples,)

get_prob¶

The method used to compute class probabilities (binary or multiclass).

Type:: callable

evaluate(y_true, y_pred, list_metrics=('AS', 'RS'))[source]¶

Evaluate the classification model using specified metrics.

Parameters:

y_true (array-like) – True target class labels.
y_pred (array-like) – Predicted class labels.
list_metrics (tuple of str, optional) – List of metrics for evaluation (default: (“AS”, “RS”)).

Returns:

Dictionary of calculated metric values.

Return type:

dict

fit(X, y)[source]¶

Fit the locally weighted classification model.

Parameters:

X (array-like, shape (n_samples, n_features)) – The training data.
y (array-like, shape (n_samples,)) – The target class labels.

Returns:

self – The fitted model.

Return type:

LwClassifier

predict(X)[source]¶

Predict class labels for the given input data.

Parameters:: X (array-like, shape (n_samples, n_features)) – The input data.
Returns:: y_pred – The predicted class labels.
Return type:: ndarray, shape (n_samples,)

predict_proba(X)[source]¶

Predict class probabilities for the given input data.

Parameters:: X (array-like, shape (n_samples, n_features)) – The input data.
Returns:: probas – The predicted class probabilities.
Return type:: ndarray, shape (n_samples, n_classes)

score(X, y)[source]¶

Compute the accuracy score for the model.

Parameters:

X (array-like, shape (n_samples, n_features)) – The input data.
y (array-like, shape (n_samples,)) – The true target class labels.

Returns:

score – The accuracy score of the predictions.

Return type:

float

scores(X, y, list_metrics=('AS', 'RS'))[source]¶

Compute evaluation metrics for the model on the given data.

Parameters:

X (array-like, shape (n_samples, n_features)) – The input data.
y (array-like, shape (n_samples,)) – The true target class labels.
list_metrics (tuple of str, optional) – List of metrics for evaluation (default: (“AS”, “RS”)).

Returns:

Dictionary of calculated metric values.

Return type:

dict

class pylwl.models.classic_lw.LwRegressor(kernel='gaussian', tau=1.0)[source]¶

Bases: BaseLW, RegressorMixin

Locally Weighted Regressor.

This class implements a locally weighted regression model using a specified kernel function and bandwidth parameter. It predicts target values by fitting a weighted linear model for each query point.

Parameters:

kernel (str or callable, optional) – The kernel function to use. If a string is provided, it should match the name of a kernel function in the kernel_module. If a callable is provided, it should accept distances and tau as arguments and return weights.
tau (float, optional) – The bandwidth parameter for the kernel function (default: 1.0).

X_¶

The training data.

Type:: ndarray, shape (n_samples, n_features)

y_¶

The target values for the training data.

Type:: ndarray, shape (n_samples,)

evaluate(y_true, y_pred, list_metrics=('MSE', 'MAE'))[source]¶

Evaluate the regression model using specified metrics.

Parameters:

y_true (array-like) – True target values.
y_pred (array-like) – Predicted target values.
list_metrics (tuple of str, optional) – List of metrics for evaluation (default: (“MSE”, “MAE”)).

Returns:

Dictionary of calculated metric values.

Return type:

dict

fit(X, y)[source]¶

Fit the locally weighted regression model.

Parameters:

X (array-like, shape (n_samples, n_features)) – The training data.
y (array-like, shape (n_samples,)) – The target values.

Returns:

self – The fitted model.

Return type:

LwRegressor

predict(X)[source]¶

Predict target values for the given input data.

Parameters:: X (array-like, shape (n_samples, n_features)) – The input data.
Returns:: y_pred – The predicted target values.
Return type:: ndarray, shape (n_samples,)

score(X, y)[source]¶

Compute the R^2 score for the model.

Parameters:

X (array-like, shape (n_samples, n_features)) – The input data.
y (array-like, shape (n_samples,)) – The true target values.

Returns:

score – The R^2 score of the predictions.

Return type:

float

scores(X, y, list_metrics=('MSE', 'MAE'))[source]¶

Compute evaluation metrics for the model on the given data.

Parameters:

X (array-like, shape (n_samples, n_features)) – The input data.
y (array-like, shape (n_samples,)) – The true target values.
list_metrics (tuple of str, optional) – List of metrics for evaluation (default: (“MSE”, “MAE”)).

Returns:

Dictionary of calculated metric values.

Return type:

dict

pylwl.models.gd_lw module¶

class pylwl.models.gd_lw.GdLwClassifier(kernel='gaussian', tau=1.0, epochs=20, optim='Adam', optim_paras=None, seed=42, verbose=True, device=None)[source]¶

Bases: BaseModel, ClassifierMixin

Gradient-Descent-based Locally Weighted Classifier.

This class implements a locally weighted classification model using gradient descent optimization. It supports binary and multiclass classification with a specified kernel function and bandwidth parameter.

Parameters:

kernel (str or callable, optional) – The kernel function to use. If a string is provided, it should match the name of a kernel function in the kernel_module. If a callable is provided, it should accept distances and tau as arguments and return weights.
tau (float, optional) – The bandwidth parameter for the kernel function (default: 1.0).
epochs (int, optional) – The number of training epochs for the local model (default: 20).
optim (str, optional) – The optimizer to use for training (default: “Adam”).
optim_paras (dict, optional) – Additional parameters for the optimizer (default: None).
seed (int, optional) – Random seed for reproducibility (default: 42).
verbose (bool, optional) – Whether to print training progress (default: True).
device (str, optional) – The device to use for training (“cpu” or “gpu”). If “gpu” is specified but not available, an error is raised (default: None).

kernel¶

The kernel function used for computing weights.

Type:: str or callable

tau¶

The bandwidth parameter for the kernel function.

Type:: float

epochs¶

The number of training epochs for the local model.

Type:: int

optim¶

The optimizer used for training.

Type:: str

optim_paras¶

Additional parameters for the optimizer.

Type:: dict

seed¶

Random seed for reproducibility.

Type:: int

verbose¶

Whether to print training progress.

Type:: bool

device¶

The device used for training (“cpu” or “cuda”).

Type:: str

X_¶

The training data.

Type:: ndarray, shape (n_samples, n_features)

y_¶

The target class labels for the training data.

Type:: ndarray, shape (n_samples,)

classes_¶

The unique class labels.

Type:: ndarray, shape (n_classes,)

is_binary_¶

Whether the classification task is binary.

Type:: bool

kernel_func_¶

The resolved kernel function (either from kernel_module or the provided callable).

Type:: callable

_trainer¶

The method used to train the local model (binary or multiclass).

Type:: callable

evaluate(y_true, y_pred, list_metrics=('AS', 'RS'))[source]¶: Evaluate the model using the specified metrics.

fit(X, y)[source]¶: Fit the model to the training data.

predict(X)[source]¶: Predict the class labels for the input samples.

predict_proba(X)[source]¶: Predict the class probabilities for the input samples.

score(X, y)[source]¶: Returns the accuracy of the model on the given data and labels.

scores(X, y, list_metrics=('AS', 'RS'))[source]¶: Compute the evaluation metrics for the model on the given data and labels.

class pylwl.models.gd_lw.GdLwRegressor(kernel='gaussian', tau=1.0, epochs=20, optim='Adam', optim_paras=None, seed=42, verbose=True, device=None)[source]¶

Bases: BaseModel, RegressorMixin

Gradient-Descent-based Locally Weighted Regressor.

This class implements a locally weighted regression model using gradient descent optimization. It predicts target values by training a local model for each query point with a specified kernel function and bandwidth parameter.

Parameters:

kernel (str or callable, optional) – The kernel function to use. If a string is provided, it should match the name of a kernel function in the kernel_module. If a callable is provided, it should accept distances and tau as arguments and return weights.
tau (float, optional) – The bandwidth parameter for the kernel function (default: 1.0).
epochs (int, optional) – The number of training epochs for the local model (default: 20).
optim (str, optional) – The optimizer to use for training (default: “Adam”).
optim_paras (dict, optional) – Additional parameters for the optimizer (default: None).
seed (int, optional) – Random seed for reproducibility (default: 42).
verbose (bool, optional) – Whether to print training progress (default: True).
device (str, optional) – The device to use for training (“cpu” or “gpu”). If “gpu” is specified but not available, an error is raised (default: None).

kernel¶

The kernel function used for computing weights.

Type:: str or callable

tau¶

The bandwidth parameter for the kernel function.

Type:: float

epochs¶

The number of training epochs for the local model.

Type:: int

optim¶

The optimizer used for training.

Type:: str

optim_paras¶

Additional parameters for the optimizer.

Type:: dict

seed¶

Random seed for reproducibility.

Type:: int

verbose¶

Whether to print training progress.

Type:: bool

device¶

The device used for training (“cpu” or “cuda”).

Type:: str

X_¶

The training data.

Type:: ndarray, shape (n_samples, n_features)

y_¶

The target values for the training data.

Type:: ndarray, shape (n_samples,)

kernel_func_¶

The resolved kernel function (either from kernel_module or the provided callable).

Type:: callable

evaluate(y_true, y_pred, list_metrics=('MSE', 'MAE'))[source]¶

Evaluate the regression model using specified metrics.

Parameters:

y_true (array-like) – True target values.
y_pred (array-like) – Predicted target values.
list_metrics (tuple of str, optional) – List of metrics for evaluation (default: (“MSE”, “MAE”)).

Returns:

Dictionary of calculated metric values.

Return type:

dict

fit(X, y)[source]¶

Fit the locally weighted regression model.

Parameters:

X (array-like, shape (n_samples, n_features)) – The training data.
y (array-like, shape (n_samples,)) – The target values.

Returns:

self – The fitted model.

Return type:

GdLwRegressor

predict(X)[source]¶

Predict target values for the given input data.

Parameters:: X (array-like, shape (n_samples, n_features)) – The input data.
Returns:: preds – The predicted target values.
Return type:: ndarray, shape (n_samples,)

score(X, y)[source]¶

Compute the R2 score for the model.

Parameters:

X (array-like, shape (n_samples, n_features)) – The input data.
y (array-like, shape (n_samples,)) – The true target values.

Returns:

score – The R2 score of the predictions.

Return type:

float

scores(X, y, list_metrics=('MSE', 'MAE'))[source]¶

Compute evaluation metrics for the model on the given data.

Parameters:

X (array-like, shape (n_samples, n_features)) – The input data.
y (array-like, shape (n_samples,)) – The true target values.
list_metrics (tuple of str, optional) – List of metrics for evaluation (default: (“MSE”, “MAE”)).

Returns:

Dictionary of calculated metric values.

Return type:

dict