def create_model():
try:
# Get opjectId, collabel, feature, algorithm and parameters
data_name = request.args.get("dataName")
class_name = request.args.get("className")
model_name = request.args.get("modelname")
col_label = int(request.args.get("label"))
col_feature_str = (request.args.get("feature")).split(",")
col_feature = []
for col in col_feature_str:
col_feature.append(int(col))
athm = request.args.get("algorithm")
athm_id = request.args.get("algorithmId")
params = json.loads(request.args.get("params"))
# print(params)
if params == {}:
params = None
test_size = 0.3
else:
test_size = float(params["testSize"])
if test_size >= 1.0 or test_size <= 0.0:
data = {"error": "0.0 < test size < 1.0"}
return data
# get data
r = API.get_class(class_name)
arr = str(r.data, "utf-8")
r_json = json.loads(arr)
data = r_json["results"]
dataFrame = pd.DataFrame(data)
if class_name == "DatasetSurveyBalance":
dataFrame = dataFrame.iloc[:, 3:]
elif class_name == "DatasetObesity":
dataFrame = dataFrame.iloc[:, 3:]
if "yearOfBirth" in list(dataFrame.columns):
del dataFrame["yearOfBirth"]
# dataFrame = dataFrame.dropna(axis="1",how = "any")
col_feature_name = np.array((dataFrame.iloc[:, col_feature]).columns)
col_feature_name_str = col_feature_name[0]
col_feature_name = list(col_feature_name)
col_feature_name.pop(0)
col_feature_name = np.array(col_feature_name)
for col in col_feature_name:
col_feature_name_str = col_feature_name_str + "," + col
col_label_name = str(
np.array(
pd.DataFrame(np.matrix(dataFrame.columns)).iloc[0, col_label]))
# get data train, test
X_train, X_test, y_train, y_test = DATA.get_data_train(
dataFrame, col_feature, col_label, test_size)
model, evalution, error, params = get_athm(athm, X_train, X_test,
y_train, y_test, params)
if error != "":
data = {"error": error}
return data
else:
# Create random id for file name
folder_model = "./upload_model"
randomId = str(uuid.uuid4())[:8]
file_name_model = (randomId + "_" + str(athm) + "_" +
str(class_name) + str(".pkl"))
pkl_filename = folder_model + "/" + file_name_model
joblib.dump(model, str(file_name_model))
custom_header = {}
custom_header[
"X-Parse-Application-Id"] = API.X_Parse_Application_Id
custom_header["X-Parse-REST-API-Key"] = API.X_Parse_REST_API_Key
custom_header["Content-Type"] = "application/x-binary"
desription = description = (
"Model " + " use " + str(athm) + " algorithm " + ". " +
"Dataset for model is " + str(data_name) +
", columns label is " + str(col_label_name) +
" and columns feature is " + str(col_feature_name))
r_upload = API.upload_model_file(
file_name_model,
model_name,
data_name,
athm_id,
params,
col_label,
col_label_name,
col_feature,
col_feature_name_str,
description,
evalution,
)
return r_upload
except:
print("[error] (createModel function app.py)")
data = {"error": "can't create model"}
return data
def create_model_system_mx():
try:
# Get opjectId, collabel, feature, algorithm and parameters
data_name = 'DatasetMX'
class_name = 'DatasetObesity'
model_name = "MODEL SYSTEM"
col_label = 7
col_feature_str = [
'0', '1', '2', '3', '4', '5', '6', '8', '9', '10', '11', '12',
'13', '14', '15', '16'
]
col_feature = []
for col in col_feature_str:
col_feature.append(int(col))
athm = 'SupportVectorMachine'
athm_id = 'ccn7ofeacm'
params = {
'C': '100000',
'degree': '3',
'gamma': '0.3',
'kernel': 'linear',
'testSize': '0.3'
}
if params == {}:
params = None
test_size = 0.3
else:
test_size = float(params["testSize"])
if test_size >= 1.0 or test_size <= 0.0:
data = {"error": "0.0 < test size < 1.0"}
return data
# get data
r = API.get_class(class_name)
arr = str(r.data, "utf-8")
r_json = json.loads(arr)
data = r_json["results"]
dataFrame = pd.DataFrame(data)
if class_name == "DatasetSurveyBalance":
dataFrame = dataFrame.iloc[:, 3:]
elif class_name == "DatasetObesity":
dataFrame = dataFrame.iloc[:, 3:]
if "yearOfBirth" in list(dataFrame.columns):
del dataFrame["yearOfBirth"]
# dataFrame = dataFrame.dropna(axis="1",how = "any")
col_feature_name = np.array((dataFrame.iloc[:, col_feature]).columns)
col_feature_name_str = col_feature_name[0]
col_feature_name = list(col_feature_name)
col_feature_name.pop(0)
col_feature_name = np.array(col_feature_name)
for col in col_feature_name:
col_feature_name_str = col_feature_name_str + "," + col
col_label_name = str(
np.array(
pd.DataFrame(np.matrix(dataFrame.columns)).iloc[0, col_label]))
# get data train, test
X_train, X_test, y_train, y_test = DATA.get_data_train(
dataFrame, col_feature, col_label, test_size)
model, evalution, error, params = get_athm(athm, X_train, X_test,
y_train, y_test, params)
if error != "":
data = {"error": error}
return data
else:
# Create random id for file name
folder_model = "./upload_model"
randomId = str(uuid.uuid4())[:8]
file_name_model = (randomId + "_" + str(athm) + "_" +
str(class_name) + str(".pkl"))
pkl_filename = folder_model + "/" + file_name_model
joblib.dump(model, str(file_name_model))
custom_header = {}
custom_header[
"X-Parse-Application-Id"] = API.X_Parse_Application_Id
custom_header["X-Parse-REST-API-Key"] = API.X_Parse_REST_API_Key
custom_header["Content-Type"] = "application/x-binary"
desription = description = (
"Model " + " use " + str(athm) + " algorithm " + ". " +
"Dataset for model is " + str(data_name) +
", columns label is " + str(col_label_name) +
" and columns feature is " + str(col_feature_name))
r_upload_2 = API.upload_model_file_system_mx(
file_name_model,
model_name,
data_name,
athm_id,
params,
col_label,
col_label_name,
col_feature,
col_feature_name_str,
description,
evalution,
)
return r_upload_2
except:
print("[error] (createModel function app.py)")
data = {"error": "can't create model"}
return data
