def test_filename_friendly_hash():
data = {
'stuff': 'stuff',
'other_stuff': 'more_stuff',
'a_datetime': datetime.datetime(2015, 1, 1),
'a_date': datetime.date(2016, 1, 1),
'a_number': 5.0
}
output = filename_friendly_hash(data)
assert isinstance(output, str)
assert re.match('^[\w]+$', output) is not None
# make sure ordering keys differently doesn't change the hash
new_output = filename_friendly_hash({
'other_stuff':
'more_stuff',
'stuff':
'stuff',
'a_datetime':
datetime.datetime(2015, 1, 1),
'a_date':
datetime.date(2016, 1, 1),
'a_number':
5.0
})
assert new_output == output
# make sure new data hashes to something different
new_output = filename_friendly_hash({'stuff': 'stuff', 'a_number': 5.0})
assert new_output != output
def test_filename_friendly_hash_stability():
nested_data = {'one': 'two', 'three': {'four': 'five', 'six': 'seven'}}
output = filename_friendly_hash(nested_data)
# 1. we want to make sure this is stable across different runs
# so hardcode an expected value
assert output == '9a844a7ebbfd821010b1c2c13f7391e6'
other_nested_data = {
'one': 'two',
'three': {
'six': 'seven',
'four': 'five'
}
}
new_output = filename_friendly_hash(other_nested_data)
assert output == new_output
def _model_hash(self, matrix_metadata, class_path, parameters):
"""Generates a unique identifier for a trained model
based on attributes of the model that together define
equivalence; in other words, if we train a second model with these
same attributes there would be no reason to keep the old one)
Args:
class_path (string): a full class path for the classifier
parameters (dict): all hyperparameters to be passed to the classifier
Returns: (string) a unique identifier
"""
unique = {
'className': class_path,
'parameters': parameters,
'project_path': self.project_path,
'training_metadata': matrix_metadata
}
return filename_friendly_hash(unique)