def __init__(self, conf: Conf):
super().__init__()
# validate python version
Validation.python_version(
Evaluator.REQUIRED_PYTHON, f"Unsupported Python version.\n"
f"Required Python {Evaluator.REQUIRED_PYTHON[0]}.{Evaluator.REQUIRED_PYTHON[1]} or higher."
)
Validation.can_read(
conf.dataset_train,
f"Training set file *must* exists and be readable. "
f"Current file: '{conf.dataset_train}'.\n"
f"Training set path (fully qualified) can be specified in conf.ini file or using Conf object."
)
Validation.can_read(
conf.dataset_test, f"Test set file *must* exists and be readable. "
f"Current file: '{conf.dataset_test}'.\n"
f"Test set path (fully qualified) can be specified in conf.ini file or using Conf object."
)
self.__LOG = LogManager.get_instance().logger(LogManager.Logger.EVAL)
self.__conf = conf
# using full dataset as training set
self.__training = Set(pd.read_csv(self.conf.dataset_train))
# load test set if it has same format as training_set.csv provided
# as example file see ./res/dataset/test_set_no_index.csv
self.__test = Set(pd.read_csv(self.conf.dataset_test))
# load test set if it has header (F1-20 and CLASS row) and index, so a test test saved using
# command pd.to_csv('/path', index=True)
# as example file see ./res/dataset/test_set_index.csv
# self.__test = Set(pd.read_csv(self.conf.dataset_test, index_col=0))
# load test set if it does not have header row (does not have F1-20 and CLASS row) and
# it is was not saved using command pd.to_csv('/path', index=True), so it has not index
# as example file see ./res/dataset/test_set_no_index_features.csv
# self.__test = Set(pd.read_csv(self.conf.dataset_test, header=None,
# names=[f"F{i}" for i in range(1, 21)] + ["CLASS"]))
# current classifiers used
self.__classifiers = {
Evaluator._MULTILAYER_PERCEPTRON:
None,
Evaluator._SUPPORT_VECTOR_MACHINE:
None,
# Evaluator._DECISION_TREE: None,
Evaluator._RANDOM_FOREST:
None,
# Evaluator._KNEAREST_NEIGHBORS: None,
# Evaluator._STOCHASTIC_GRADIENT_DESCENT: None,
Evaluator._ADA_BOOST:
None,
# Evaluator._NAIVE_BAYES: None,
# Evaluator._KMEANS: None
}
def __check_permissions(self):
"""
Check permissions on directories before performing the operations
:raise ValueError if input directory is equal to output directory
:raise NotADirectoryError
:raise PermissionError
:raise LinksError
"""
sources = self.__dispatcher_config.dispatcher_sources
destinations = self.__dispatcher_config.dispatcher_destinations
for source in sources:
Validation.is_dir(
sources[source],
f"Missing input directory '{sources[source]}'"
)
Validation.can_read(
sources[source],
f"Missing read permission on '{sources[source]}'"
)
Validation.can_write(
sources[source],
f"Missing write permission on '{sources[source]}'"
)
for destination in destinations:
try:
Validation.is_dir_writeable(
destinations[destination],
f"Directory '{destinations[destination]}' *must* exists and be writable"
)
except NotADirectoryError:
parent_directory = Path(destinations[destination]).parent
Validation.can_write(
parent_directory,
f"Missing write permission on '{parent_directory}'"
)
FileObserver.__LOG.info(f"Creating missing destination directory '{destinations[destination]}'")
# create if not exists
Path(destinations[destination]).mkdir(parents=True, exist_ok=True)
for source in sources:
Validation.are_symlinks(
sources[source],
destinations[destination],
f"Input ('{sources[source]}') and output ('{destinations[destination]}') directory can not be the same (or symlinks)"
)