def modellist(name):
message = "Success"
code = 200
result = []
try:
service = projectmgr.GetService(name,
constants.ServiceTypes.MachineLearning)
servicedata = json.loads(service.servicedata)
modeltype = servicedata["model_type"]
if modeltype == "mlp":
models = projectmgr.GetDeepModels(
name, constants.ServiceTypes.MachineLearning)
for m in models:
result.append({
"name": m.modelname,
"modifiedon": m.modifiedon
})
elif modeltype == "general":
models = scikitlearn.getModels()
for m in models:
result.append({"name": m, "modifiedon": datetime.utcnow()})
except Exception as e:
code = 500
message = str(e)
return jsonify({"statuscode": code, "message": message, "result": result})
def Validate(id, name):
results = {}
status = "Completed"
message = "Completed"
try:
srvjson = json.loads(projectmgr.GetService(name, "ml").servicedata)
model_type = srvjson["model_type"]
pipeline.init(pipeline, name, model_type, id)
pipelinejson = pipeline.getPipelineData()
pipeline.Run()
for p in pipelinejson:
if p["module"] == "return_result":
mlist = p["input"]["module_output"]
for m in mlist:
r = pipeline.Output(m)
results[m] = json.loads(r)
except Exception as e:
results["message"] = pipeline.lastpipeline + ": " + str(e)
message = str(e)
status = "Error"
projectmgr.EndJob(id, status, message, json.dumps(results))
def getServiceJson(srvname, srvpath):
result = {}
service = projectmgr.GetService(srvname, srvpath)
if not service is None:
result = json.loads(service.servicedata)
return result
def apilistwithname(srvtype, srvname):
message = "Success"
code = 200
result = []
try:
service = projectmgr.GetService(srvtype, srvname)
if service is None:
raise Exception("Service API not found")
result = json.loads(service.servicedata)
except Exception as e:
code = 500
message = str(e)
return jsonify({"statuscode": code, "message": message, "result": result})
def predict(name, text):
botfolder = "./data/__chatbot/" + name
service = projectmgr.GetService(name, constants.ServiceTypes.ChatBot)
botjson = json.loads(service.servicedata)
bot = getBot(name)
response = bot.get_response(text.lower())
result = {
"confidence": response.confidence,
"response_text": response.text
}
if float(response.confidence) < float(botjson["threshold"]):
result = {
"confidence": response.confidence,
"response_text": botjson["default_response"]
}
return result
def predict(name):
message = "Success"
code = 200
result = []
try:
start = datetime.now()
data = json.loads(request.data)
service = projectmgr.GetService(name,
constants.ServiceTypes.MachineLearning)
servicejson = json.loads(service.servicedata)
savePrediction = False
if "save_prediction" in data:
savePrediction = data['save_prediction']
result = {}
if servicejson["data_format"] == "image":
testfile = data['imagepath']
elif servicejson["data_format"] == "csv":
testfile = data['testfile']
pipeline.init(pipeline, name, servicejson["model_type"])
predictions = pipeline.Predict(testfile, savePrediction)
predictions = json.loads(predictions)
if servicejson["data_format"] == "csv":
result = predictions["0"]
else:
result = predictions
logmgr.LogPredSuccess(name, constants.ServiceTypes.MachineLearning,
start)
except Exception as e:
code = 500
message = str(e)
logmgr.LogPredError(name, constants.ServiceTypes.MachineLearning,
start, message)
return jsonify({"statuscode": code, "message": message, "result": result})