def add_experiment(name: Text, type: Text, train_data_filename: Text,
test_data_filename: Text, train_data, test_data):
experiment = Experiment(name=name, type=type)
db.session.add(experiment)
db.session.commit()
experiment_locator = Locator(experiment.id,
train_data_filename=train_data_filename,
test_data_filename=test_data_filename)
with open(
os.path.join(experiment_locator.get_train_data_dir(),
train_data_filename), 'wb') as copied_train_data:
for line in train_data.readlines():
copied_train_data.write(line)
with open(
os.path.join(experiment_locator.get_test_data_dir(),
test_data_filename), 'wb') as copied_test_data:
for line in test_data.readlines():
copied_test_data.write(line)
experiment.train_data = train_data_filename
experiment.test_data = test_data_filename
db.session.commit()
return experiment
def delete_single_experiment(id):
try:
experiment = Experiment.query.filter_by(id=id).first()
if not experiment:
response_object = {
'status': 'fail',
'message': f'Experiment id {id} Not Found!'
}
return jsonify(response_object), 404
else:
db.session.delete(experiment)
db.session.commit()
Locator.delete_experiment_folders(id)
response_object = {
'status': 'success',
'message': f'Experiment id {id} Deleted!'
}
return jsonify(response_object), 200
except Exception as e:
response_object = {
'status': 'fail',
'message': "Internal Server Error",
'error': str(e)
}
return jsonify(response_object), 500
def add_experiment():
"""Add experiment to database"""
try:
data, status_code = add_experiment_form_valid(request)
if data != "form_valid":
return data, status_code
name = request.form['name']
type = request.form['type']
experiment = Experiment.query.filter_by(name=name).first()
if experiment:
response_object = {
'status':
'fail',
'message':
f'Experiment with name: {name} exists, Please enter unique experiment name'
}
return jsonify(response_object), 409
train_data_file = request.files['train_data']
train_data_filename = secure_filename(train_data_file.filename)
test_data_file = request.files['test_data']
test_data_filename = secure_filename(test_data_file.filename)
experiment = Experiment(name=name, type=type)
db.session.add(experiment)
db.session.commit()
experiment_locator = Locator(experiment.id,
train_data_filename=train_data_filename,
test_data_filename=test_data_filename)
train_data_file.save(experiment_locator.get_train_data_file_path())
test_data_file.save(experiment_locator.get_test_data_file_path())
experiment.train_data = train_data_filename
experiment.test_data = test_data_filename
db.session.commit()
response_object = {
'status': 'success',
'message': f'Experiment {name} added!',
'id': experiment.id
}
return jsonify(response_object), 201
except Exception as e:
print(str(e))
response_object = {
'status': 'fail',
'message': "Internal Server Error",
'error': str(e)
}
return jsonify(response_object), 500
def predict(cls, experiment, sample):
name = experiment.name
exp_locator = Locator(experiment.id, experiment.train_data, experiment.test_data)
if 'LR_' in name:
lr_dataset_maker = LRPimaIndiansDatasetMaker(exp_locator)
X = lr_dataset_maker.make_one_sample(sample)
logistic_regression = DataLoader.load(exp_locator.get_model_file_path())
prediction = logistic_regression.predict(X)
prediction = prediction.tolist()
return prediction
def update_single_experiment(id):
"""Update Single Experiment"""
try:
experiment = Experiment.query.filter_by(id=id).first()
if not experiment:
response_object = {
'status': 'fail',
'message': f'Experiment id {id} Not Found!'
}
return jsonify(response_object), 404
name = request.form['name']
type = request.form['type']
result = request.form['result']
start_date = request.form['start_date']
start_date = datetime.strptime(start_date, '%Y-%m-%d %H:%M:%S.%f')
train_data_file = request.files['train_data']
train_data_filename = secure_filename(train_data_file.filename)
test_data_file = request.files['test_data']
test_data_filename = secure_filename(test_data_file.filename)
experiment_locator = Locator(experiment.id,
train_data_filename=train_data_filename,
test_data_filename=test_data_filename)
train_data_file.save(experiment_locator.get_train_data_file_path())
test_data_file.save(experiment_locator.get_test_data_file_path())
experiment.name = name
experiment.type = type
experiment.result = result
experiment.start_date = start_date
experiment.train_data = train_data_filename
experiment.test_data = test_data_filename
db.session.commit()
response_object = {
'status': 'success',
'message': f'Experiment id {id} Updated!'
}
return jsonify(response_object), 200
except Exception as e:
response_object = {
'status': 'fail',
'message': "Internal Server Error",
'error': str(e)
}
return jsonify(response_object), 500
def train(cls, experiment):
name = experiment.name
exp_locator = Locator(experiment.id, experiment.train_data, experiment.test_data)
if 'LR_' in name:
lr_dataset_maker = LRPimaIndiansDatasetMaker(exp_locator)
X, y = lr_dataset_maker.make_train_dataset()
hyperparams = {'penalty': 'l2'}
logistic_regression = Logistic_Regression(hyperparams=hyperparams)
result = logistic_regression.train(X, y)
DataLoader.save(file_object=logistic_regression,
file_path=os.path.join(exp_locator.get_model_dir(), 'model.pkl'))
experiment.result = json.dumps([result])
return experiment
raise Exception("No valid Experiment Name to train Experiment")
def test(cls, experiment):
name = experiment.name
exp_locator = Locator(experiment.id, experiment.train_data, experiment.test_data)
if 'LR_' in name:
lr_dataset_maker = LRPimaIndiansDatasetMaker(exp_locator)
X, y = lr_dataset_maker.make_test_dataset()
logistic_regression = DataLoader.load(exp_locator.get_model_file_path())
test_result = logistic_regression.test(X, y)
exp_result = json.loads(experiment.result)
exp_result.append(test_result)
experiment.result = json.dumps(exp_result)
return experiment
raise Exception("No valid Experiment Name to train Experiment")
def test_create_experiment_folders(self):
"""Ensure create experiment folder behaves properly"""
locator = Locator(id=1, train_data_filename='pima_indians.csv', test_data_filename='pima_indians.csv')
self.assertTrue(os.path.exists(locator.train_data_dir))
self.assertTrue(os.path.exists(locator.test_data_dir))
self.assertTrue(os.path.exists(locator.model_dir))
self.assertTrue(os.path.exists(locator.train_result_dir))
self.assertTrue(os.path.exists(locator.test_result_dir))
def clean_up_folders(ids):
_ = [Locator.delete_experiment_folders(i) for i in ids]
def is_experiment_trained(cls, experiment):
exp_locator = Locator(experiment.id, experiment.train_data, experiment.test_data)
if os.path.exists(exp_locator.get_model_file_path()):
return True
else:
return False
def test_get_test_result_dir(self):
"""Ensure get_train_data_dir behaves properly"""
locator = Locator(id=1, train_data_filename='pima_indians.csv', test_data_filename='pima_indians.csv')
self.assertIn('/experiments/1/result/test', locator.get_test_result_dir())