def update_sensor(sensor_id):
'''Updates sensor information
Updates existing virtual sensor information and returns updated information.
Args as data:
{
"_id": Object ID of an existing virtual sensor
"name": A name of a virtual sensor
"user_id": ID of a user who creates this sensor
"labels": An array of strings denoting labels
"inputs": An array of UUIDs of real sensors used as inputs for a classifier
"sensor_uuid": A sensor UUID of this virtual sensor in BD (can pass blank)
"description": A description of this virtual sensor
}
Returns:
{
"url": A URL of the HTTP call
"method": "PUT"
"result": ok or error
"ret":{
"_id": A virtual sensor's object ID
"name": A sensor name
"sensor_uuid": A UUID of a sensor if it's registered in BD
"user_id": An ID of a user who created this sensor
"labels": An array of labels
"inputs": An array of sensor UUIDs in BD used as inputs for a classifier
"description": A description of a sensor
}
}
'''
user_id = 'default'
dic = {
'uri':request.url,
'method':request.method
}
sensor = MLSensor(request.get_json())
result = database_manager.update_sensor(sensor)
if result is not None:
dic['result'] = 'ok'
dic['ret'] = database_manager.sensor(sensor._id, user_id).to_dictionary()
else:
dic['result'] = 'error'
return jsonString(dic)
def predict(sensor_id, user_id, end_time=None):
'''Makes a prediction with a virtual sensor
Makes a prediciton using a pre-trained classifer with timeseries data in a range
[end time - sampling period, end time]. sampling period is an average of sampling
durations in a traing set, which is stored as a part of a classifier.
When end_time is not specifiec, current time is used as end_time.
This function generated a sample on timestamps passed to this function. Then,
makes a prediction using a pre-trained classifier.
Actual feature extraction and prediction are implemented in a classifier class.
Check random_forest.py for current the implementation of the current classifier.
Args:
sensor_id: An object ID of a virtual sensor.
user_id: A user ID of a user who own the virtual sensor.
Returns:
cls_result: An instance of a container class MLClassifierResult
'''
clf_result = MLClassifierResult()
classifier = db_manager.classifier(sensor_id, user_id)
if classifier is None: # classifier not found in the database
clf_result.result = 'error'
clf_result.message = 'A classifier for the sensor not found.'
return clf_result
sensor = db_manager.sensor(sensor_id, user_id)
if end_time is not None:
end_time = float(end_time)
timeseries = db_manager.timeseries_for_inputs(
sensor.inputs, end_time - classifier.sampling_period, end_time)
else:
timeseries = db_manager.latest_timeseries_for_inputs(
sensor.inputs, classifier.sampling_period)
prediction = classifier.predict(timeseries) # make prediction
if prediction is None: # cannot make a prediction
clf_result.result = 'error'
clf_result.message = 'A classification error occurred.'
return clf_result
clf_result.prediction = prediction
print prediction
return clf_result
def predict(sensor_id, user_id, end_time=None):
'''Makes a prediction with a virtual sensor
Makes a prediciton using a pre-trained classifer with timeseries data in a range
[end time - sampling period, end time]. sampling period is an average of sampling
durations in a traing set, which is stored as a part of a classifier.
When end_time is not specifiec, current time is used as end_time.
This function generated a sample on timestamps passed to this function. Then,
makes a prediction using a pre-trained classifier.
Actual feature extraction and prediction are implemented in a classifier class.
Check random_forest.py for current the implementation of the current classifier.
Args:
sensor_id: An object ID of a virtual sensor.
user_id: A user ID of a user who own the virtual sensor.
Returns:
cls_result: An instance of a container class MLClassifierResult
'''
clf_result = MLClassifierResult()
classifier = db_manager.classifier(sensor_id, user_id)
if classifier is None: # classifier not found in the database
clf_result.result = 'error'
clf_result.message = 'A classifier for the sensor not found.'
return clf_result
sensor = db_manager.sensor(sensor_id, user_id)
if end_time is not None:
end_time = float(end_time)
timeseries = db_manager.timeseries_for_inputs(sensor.inputs, end_time-classifier.sampling_period, end_time)
else:
timeseries = db_manager.latest_timeseries_for_inputs(sensor.inputs, classifier.sampling_period)
prediction = classifier.predict(timeseries) # make prediction
if prediction is None: # cannot make a prediction
clf_result.result = 'error'
clf_result.message = 'A classification error occurred.'
return clf_result
clf_result.prediction = prediction
print prediction
return clf_result
def update_sensor(sensor_id):
'''Updates sensor information
Updates existing virtual sensor information and returns updated information.
Args as data:
{
"_id": Object ID of an existing virtual sensor
"name": A name of a virtual sensor
"user_id": ID of a user who creates this sensor
"labels": An array of strings denoting labels
"inputs": An array of UUIDs of real sensors used as inputs for a classifier
"sensor_uuid": A sensor UUID of this virtual sensor in BD (can pass blank)
"description": A description of this virtual sensor
}
Returns:
{
"url": A URL of the HTTP call
"method": "PUT"
"result": ok or error
"ret":{
"_id": A virtual sensor's object ID
"name": A sensor name
"sensor_uuid": A UUID of a sensor if it's registered in BD
"user_id": An ID of a user who created this sensor
"labels": An array of labels
"inputs": An array of sensor UUIDs in BD used as inputs for a classifier
"description": A description of a sensor
}
}
'''
user_id = 'default'
dic = {'uri': request.url, 'method': request.method}
sensor = MLSensor(request.get_json())
result = database_manager.update_sensor(sensor)
if result is not None:
dic['result'] = 'ok'
dic['ret'] = database_manager.sensor(sensor._id,
user_id).to_dictionary()
else:
dic['result'] = 'error'
return jsonString(dic)
def get_sensor(sensor_id):
'''Returns informatoin about a virtual sensor
Returns an object containing information about a virtual sensor.
Args as a part of URL:
<sensor_id>: A sensor's object ID.
Returns:
{
"url": A URL of the HTTP call
"method": "GET"
"result": ok or error
"ret":{
"_id": A virtual sensor's object ID
"name": A sensor name
"sensor_uuid": A UUID of a sensor if it's registerd in BD
"user_id": An ID of a user who created this sensor
"labels": An array of labels
"inputs": An array of sensor UUIDs in BD used as inputs for a classifier
"description": A description of a sensor
}
}
'''
user_id = 'default'
sensor = database_manager.sensor(sensor_id, user_id)
dic = {
'url':request.url,
'method':request.method,
}
if sensor is not None:
dic['ret'] = sensor.to_dictionary()
dic['result'] = 'ok'
else:
dic['result'] = 'error'
return jsonString(dic)
def get_sensor(sensor_id):
'''Returns informatoin about a virtual sensor
Returns an object containing information about a virtual sensor.
Args as a part of URL:
<sensor_id>: A sensor's object ID.
Returns:
{
"url": A URL of the HTTP call
"method": "GET"
"result": ok or error
"ret":{
"_id": A virtual sensor's object ID
"name": A sensor name
"sensor_uuid": A UUID of a sensor if it's registerd in BD
"user_id": An ID of a user who created this sensor
"labels": An array of labels
"inputs": An array of sensor UUIDs in BD used as inputs for a classifier
"description": A description of a sensor
}
}
'''
user_id = 'default'
sensor = database_manager.sensor(sensor_id, user_id)
dic = {
'url': request.url,
'method': request.method,
}
if sensor is not None:
dic['ret'] = sensor.to_dictionary()
dic['result'] = 'ok'
else:
dic['result'] = 'error'
return jsonString(dic)