def create_model():
database_url = os.environ[DATABASE_URL]
database_replica_set = os.environ[DATABASE_REPLICA_SET]
database_name = os.environ[DATABASE_NAME]
train_filename = request.json[TRAINING_FILENAME]
test_filename = request.json[TEST_FILENAME]
classifiers_name = request.json[CLASSIFIERS_NAME]
database = Database(
database_url,
database_replica_set,
os.environ[DATABASE_PORT],
database_name,
)
request_validator = UserRequest(database)
request_errors = analyse_request_errors(
request_validator,
train_filename,
test_filename,
classifiers_name)
if request_errors is not None:
return request_errors
database_url_training = Database.collection_database_url(
database_url,
database_name,
train_filename,
database_replica_set,
)
database_url_test = Database.collection_database_url(
database_url,
database_name,
test_filename,
database_replica_set,
)
metadata_creator = Metadata(database, train_filename, test_filename)
model_builder = Model(database,
metadata_creator,
database_url_training,
database_url_test)
model_builder.build(
request.json[MODELING_CODE_NAME],
classifiers_name
)
return (
jsonify({
MESSAGE_RESULT:
create_prediction_files_uri(
classifiers_name,
test_filename)}),
HTTP_STATUS_CODE_SUCCESS_CREATED,
)
def predictions_filename_validator(self, test_filename, classifier_list):
filenames = self.database.get_filenames()
for classifier_name in classifier_list:
prediction_filename = Model.create_prediction_filename(
test_filename, classifier_name)
if prediction_filename in filenames:
raise Exception(self.MESSAGE_INVALID_PREDICTION_NAME)
def create_prediction_files_uri(classifiers_list, test_filename):
classifiers_uri = []
for classifier in classifiers_list:
classifiers_uri.append(
MICROSERVICE_URI_GET +
Model.create_prediction_filename(test_filename,
classifier) +
MICROSERVICE_URI_GET_PARAMS)
return classifiers_uri
def create_file(self, classifier_name):
metadata = self.metadata_document.copy()
metadata["classifier"] = classifier_name
metadata["datasetName"] = \
Model.create_prediction_filename(
self.test_filename,
classifier_name)
self.database_connector.delete_file(metadata["datasetName"])
self.database_connector.insert_one_in_file(metadata["datasetName"],
metadata)
return metadata