def insert_single_measurement():
"""
Route for adding a single measurement, this endpoint could be used by external sensors, devices, etc. for adding
to the database
:return: JSON
"""
REQUIRED_PARAMS = [
'date_time', 'longitude', 'latitude', 'pollutant', 'pollution_value'
]
body = request.json
missing_param = Helpers.are_params_missing(
body, REQUIRED_PARAMS) # assume no param is missing
if missing_param:
return response_failure(Errors.MISSING_PARAM.value)
is_successful, err = DatasetsApi.insert_single_measurement(body)
# Assume instance has not been inserted
response_msg = {'instance_inserted': err}
if is_successful:
response_msg = {'instance_inserted': True}
return response_success(response_msg)
return response_failure(response_msg)
def get_model_types():
"""
Route for getting all supported models' types, currently only CNN, FullGP, SparseGP
:return: JSON
"""
model_types = ModelApi.get_model_types()
if not isinstance(model_types, list):
return response_failure(Errors.NO_DATA.value)
return response_success(model_types)
def get_all_models():
"""
Route for getting all models' names and types only from the DB
:return: JSON
"""
models = ModelApi.get_all_models()
if models is None:
return response_failure(models)
return response_success(models)
def get_models(model_type):
"""
Route for getting models that are of given model type (CNN, GP, SparseGP up to date)
:param model_type:
:return: JSON
"""
models, err = ModelApi.get_models_by_type(model_type)
if models is not None:
return response_success(models)
return response_failure(err)
def get_model_params(name):
"""
Route for getting model parameters of a given model
:param name: str - name of the model
:return: JSON
"""
model = ModelApi.get_model_params(name)
if model is not None:
return response_success(model)
return response_failure(Errors.NO_SUCH_MODEL.value)
def make_predictions():
"""
Function for making predictions based on the datetime and locations given for a single pollutant, type of model
is also required (CNN/FullGP/SparseGP)
:return: JSON
"""
REQUIRED_PARAMS = ['name', 'range', 'locations', 'pollutant']
body = request.json
missing_param = Helpers.are_params_missing(
body, REQUIRED_PARAMS) # assume no param is missing
if missing_param:
return response_failure(Errors.MISSING_PARAM.value)
if not isinstance(body['range'], dict):
return response_failure(Errors.WRONG_PARAM.value)
result = ConcurrentTaskApi.concurrent_make_predictions.delay(body)
return response_success(True)
def get_pollution_levels():
"""
Function for getting the pollution levels in a given time range for given locations
This method is POST due to the high number of parameters included
:return: JSON
"""
REQUIRED_PARAMS = ['range', 'locations', 'pollutant']
body = request.json
missing_param = Helpers.are_params_missing(
body, REQUIRED_PARAMS) # assume no param is missing
if missing_param:
return response_failure(Errors.MISSING_PARAM.value)
if not isinstance(body['range'], dict):
return response_failure(Errors.WRONG_PARAM.value)
dataset = DatasetsApi.get_dataset(body, use_dataframe=False)
return response_success(dataset)
def create_model(name):
"""
Function to create a model from given type and train it on a dataset with given parameters
:param name: unique name of the created model
:return: JSON
"""
REQUIRED_PARAMS = ['type', 'range', 'locations', 'pollutant']
body = request.json
missing_param = Helpers.are_params_missing(
body, REQUIRED_PARAMS) # assume no param is missing
if missing_param:
return response_failure(Errors.MISSING_PARAM.value)
if not isinstance(body['range'], dict):
return response_failure(Errors.WRONG_PARAM.value)
result = ConcurrentTaskApi.concurrent_create_model.delay(name, body)
return response_success(True)
def predict_single_instance():
"""
Route for creating a new arbitrary instance and making prediction on its
:return: JSON
"""
REQUIRED_PARAMS = [
'name', 'date_time', 'longitude', 'latitude', 'pollutant'
]
body = request.json
missing_param = Helpers.are_params_missing(body, REQUIRED_PARAMS)
if missing_param:
return response_failure(Errors.MISSING_PARAM.value)
is_successful, predictions = ModelApi.make_single_prediction(body)
if not is_successful:
# TODO update error message
return response_failure('Something went wrong')
return response_success(predictions)
def insert_dataset():
"""
Route for adding a whole dataset to the database, not preferable to use it with large datasets, dataset must be in
CSV
:return: JSON
"""
pollutant = None
weather = None
dataset_metadata = None
try:
pollutant = request.files['pollutant']
weather = request.files['weather']
dataset_metadata = request.files['metadata']
except:
print('no such file')
if pollutant is None:
return response_failure({'file': 'pollutant'})
if weather is None:
return response_failure({'file': 'weather'})
if dataset_metadata is None:
return response_failure({'file': 'metadata'})
files = {
'pollutant': pollutant,
'weather': weather,
'metadata': dataset_metadata
}
result, err = DatasetsApi.insert_dataset(files)
if err:
return response_failure(err)
return response_success('inserted')
def train_model(model_name):
"""
Route for training a model
:param model_name: str - model's name
:return: JSON
"""
REQUIRED_PARAMS = ['range', 'locations', 'pollutant']
body = request.json
are_params_missing = Helpers.are_params_missing(body, REQUIRED_PARAMS)
if are_params_missing:
return response_failure(Errors.MISSING_PARAM.value)
if not isinstance(body['range'], dict):
return response_failure(Errors.WRONG_PARAM.value)
if not (isinstance(body['locations'], dict)
or isinstance(body['locations'], list)):
return response_failure(Errors.WRONG_PARAM.value)
ConcurrentTaskApi.concurrent_train_model.delay(model_name, body)
return response_success(True)
