bayeso.bo.bo_w_trees

It defines a class of Bayesian optimization with tree-based surrogate models.

class bayeso.bo.bo_w_trees.BOwTrees(range_X: ndarray, str_surrogate: str = 'rf', str_acq: str = 'ei', normalize_Y: bool = True, str_optimizer_method_bo: str = 'random_search', str_exp: str = None, debug: bool = False)

Bases: BaseBO

It is a Bayesian optimization class with tree-based surrogate models.

Parameters:

  • range_X (numpy.ndarray) – a search space. Shape: (d, 2).

  • str_surrogate (str., optional) – the name of surrogate model.

  • str_acq (str., optional) – the name of acquisition function.

  • normalize_Y (bool., optional) – flag for normalizing outputs.

  • str_optimizer_method_bo (str., optional) – the name of optimization method for Bayesian optimization.

  • str_exp (str., optional) – the name of experiment.

  • debug (bool., optional) – flag for printing log messages.

_abc_impl = <_abc._abc_data object>
compute_acquisitions(X: ndarray, X_train: ndarray, Y_train: ndarray, trees: List) → ndarray

It computes acquisition function values over ‘X’, where X_train and Y_train are given.

Parameters:

  • X (numpy.ndarray) – inputs. Shape: (l, d).

  • X_train (numpy.ndarray) – inputs. Shape: (n, d).

  • Y_train (numpy.ndarray) – outputs. Shape: (n, 1).

  • trees (list) – list of trees.

Returns:

acquisition function values over X. Shape: (l, ).

Return type:

numpy.ndarray

Raises:

AssertionError

compute_posteriors(X: ndarray, trees: List) → ndarray

It returns posterior mean and standard deviation functions over X.

Parameters:

  • X (numpy.ndarray) – inputs to test. Shape: (l, d).

  • trees (list) – list of trees.

Returns:

posterior mean and standard deviation functions over X. Shape: ((l, ), (l, )).

Return type:

(numpy.ndarray, numpy.ndarray)

Raises:

AssertionError

get_trees(X_train, Y_train, num_trees=100, depth_max=5, size_min_leaf=1)

It returns a list of trees.

Parameters:

  • X_train (numpy.ndarray) – inputs. Shape: (n, d).

  • Y_train (numpy.ndarray) – outputs. Shape: (n, 1).

  • num_trees (int., optional) – the number of trees.

  • depth_max (int., optional) – maximum depth.

  • size_min_leaf (int., optional) – minimum size of leaves.

Returns:

list of trees.

Return type:

list

Raises:

AssertionError

optimize(X_train: ndarray, Y_train: ndarray, str_sampling_method: str = 'uniform', num_samples: int = 5000, seed: int = None) → Tuple[ndarray, dict]

It computes acquired example, candidates of acquired examples, acquisition function values for the candidates, covariance matrix, inverse matrix of the covariance matrix, hyperparameters optimized, and execution times.

Parameters:

  • X_train (numpy.ndarray) – inputs. Shape: (n, d).

  • Y_train (numpy.ndarray) – outputs. Shape: (n, 1).

  • str_sampling_method (str., optional) – the name of sampling method for acquisition function optimization.

  • num_samples (int., optional) – the number of samples.

  • seed (int., optional) – a random seed.

Returns:

acquired example and dictionary of information. Shape: ((d, ), dict.).

Return type:

(numpy.ndarray, dict.)

Raises:

AssertionError