lightonml.utils¶
This file contains some utils function to deal with data and load and save models.

class
OptionalProfiler
(condition)[source]¶ Bases:
object
Wrapper to profile functions.
Parameters: condition (bool,) – switch to return the decorated function

cast_01_to_uint8
(X)[source]¶ Casts binary data to uint8.
Parameters: X (np.ndarray,) – input data. Returns: X_uint8 – input data in uint8. Return type: np.ndarray,

get_ml_data_dir_path
()[source]¶ Get the data directory folder from the JSON config file.
Returns: Return type: pathlib.Path, location of the data folder.

load_data_from_numpy_archive
(path_to_file)[source]¶ Loads data from NumPy archive.
Parameters: path_to_file (str,) – path to the numpy archive to load. Returns:  (X_train, y_train) (tuple of np.ndarray,) – train set.
 (X_test, y_test) (tuple of np.ndarray,) – test set.

load_model
(model_path)[source]¶ Loads the model from a pickle file.
Parameters: model_path (str,) – path for the pickle file of the model. Returns: model – instance of the model. Return type: BaseEstimator, RegressorMixin or TransformerMixin and children,

save_model
(model, model_name, model_path='models')[source]¶ Saves a model in a pickle file.
Parameters:

select_subset
(X, y, classes=range(0, 10), ratio=1, random_state=None)[source]¶ Selects a subset of a dataset.
Parameters:  X (2D np.ndarray,) – input data.
 y (np.ndarray,) – targets.
 classes (list or np.ndarray,) – number of classes in the dataset.
 ratio (float,) – controls the ratio between examples.
 random_state (int, RandomState instance or None, optional, defaults to None,) – controls the pseudo random number generator used to subsample the dataset.
Returns:  X (np.ndarray,) – subsampled data.
 y (np.ndarray,) – subsampled targets.