def testClassify(self):
"""Test PredictionService.Classify implementation."""
model_path = self._GetSavedModelBundlePath()
atexit.register(self.TerminateProcs)
model_server_address = self.RunServer(PickUnusedPort(), 'default',
model_path)
print 'Sending Classify request...'
# Prepare request
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'default'
request.model_spec.signature_name = 'classify_x_to_y'
example = request.input.example_list.examples.add()
example.features.feature['x'].float_list.value.extend([2.0])
# Send request
host, port = model_server_address.split(':')
channel = implementations.insecure_channel(host, int(port))
stub = prediction_service_pb2.beta_create_PredictionService_stub(
channel)
result = stub.Classify(request, RPC_TIMEOUT) # 5 secs timeout
# Verify response
self.assertEquals(1, len(result.result.classifications))
self.assertEquals(1, len(result.result.classifications[0].classes))
expected_output = 3.0
self.assertEquals(expected_output,
result.result.classifications[0].classes[0].score)
def testClassify(self):
"""Test PredictionService.Classify implementation."""
model_path = self._GetSavedModelBundlePath()
model_server_address = TensorflowModelServerTest.RunServer(
'default', model_path)[1]
print('Sending Classify request...')
# Prepare request
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'default'
request.model_spec.signature_name = 'classify_x_to_y'
example = request.input.example_list.examples.add()
example.features.feature['x'].float_list.value.extend([2.0])
# Send request
result = self._MakeStub(model_server_address).Classify(
request, RPC_TIMEOUT)
# Verify response
self.assertEqual(1, len(result.result.classifications))
self.assertEqual(1, len(result.result.classifications[0].classes))
expected_output = 3.0
self.assertEqual(expected_output,
result.result.classifications[0].classes[0].score)
self._VerifyModelSpec(result.model_spec, request.model_spec.name,
request.model_spec.signature_name,
self._GetModelVersion(model_path))
def Request(text):
example = inputs.BuildTextExample(text_utils.TokenizeText(text))
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'default'
request.model_spec.signature_name = FLAGS.signature_def
request.input.example_list.examples.extend([example])
return request
def _BuildClassificationRequests(self, signature_name: Text):
for example in self.examples:
request = classification_pb2.ClassificationRequest()
request.model_spec.name = self._model_name
request.model_spec.signature_name = signature_name
request.input.example_list.examples.append(example)
yield request
def main(_):
host, port = FLAGS.server.split(':')
channel = implementations.insecure_channel(host, int(port))
stub = prediction_service_pb2.beta_create_PredictionService_stub(channel)
# Send request
# See prediction_service.proto for gRPC request/response details.
sample_image1 = iR2G.img2gs_res(test_image,28,28)
new_samples = np.array([sample_image1], dtype=np.float32)
image = new_samples.reshape(784,)
print(image.shape)
#f = open(test_image, 'rb')
#data = f.read()
# Feature_dict
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'default'
request.model_spec.signature_name = 'serving_default'
example = request.input.example_list.examples.add()
example.features.feature['x'].float_list.value.extend(image.astype(float))
result = stub.Classify(request, 10.0) # 10 secs timeout
print(result)
def test_pb_request():
request = classification_pb2.ClassificationRequest()
request.model_spec.name = "generic_model"
request.model_spec.signature_name = DEFAULT_SERVING_SIGNATURE_DEF_KEY
example = request.input.example_list.examples.add()
data = [4.9, 2.5, 4.5, 1.7]
example.features.feature[PREDICT_INPUTS].float_list.value.extend(data)
serialized_output = requests.post("http://localhost:8080/invocations",
data=request.SerializeToString(),
headers={
'Content-type':
'application/octet-stream',
'Accept': 'application/octet-stream'
}).content
classification_response = classification_pb2.ClassificationResponse()
classification_response.ParseFromString(serialized_output)
classifications_classes = classification_response.result.classifications[
0].classes
assert len(classifications_classes) == 3
for c in classifications_classes:
assert c.score < 1
assert c.score > 0
def classify(body):
prediction_service_stub = _get_prediction_service_stub()
request = ParseDict(body, classification.ClassificationRequest())
result = prediction_service_stub.Classify(request)
return MessageToDict(result,
preserving_proto_field_name=True,
including_default_value_fields=True)
def call_servo(examples, serving_bundle):
"""Send an RPC request to the Servomatic prediction service.
Args:
examples: A list of tf.train.Examples that matches the model spec.
serving_bundle: A `ServingBundle` object that contains the information to
make the serving request.
Returns:
A ClassificationResponse or RegressionResponse proto.
"""
parsed_url = urlparse('http://' + serving_bundle.inference_address)
channel = implementations.insecure_channel(parsed_url.hostname,
parsed_url.port)
stub = prediction_service_pb2.beta_create_PredictionService_stub(channel)
if serving_bundle.model_type == 'classification':
request = classification_pb2.ClassificationRequest()
else:
request = regression_pb2.RegressionRequest()
request.model_spec.name = serving_bundle.model_name
if serving_bundle.model_version is not None:
request.model_spec.version.value = serving_bundle.model_version
if serving_bundle.signature is not None:
request.model_spec.signature_name = serving_bundle.signature
request.input.example_list.examples.extend(examples)
if serving_bundle.model_type == 'classification':
return stub.Classify(request, 30.0) # 30 secs timeout
else:
return stub.Regress(request, 30.0) # 30 secs timeout
def main():
channel = implementations.insecure_channel(
'10.211.44.8', 8500) #the ip and port of your server host
stub = prediction_service_pb2.beta_create_PredictionService_stub(channel)
# the test samples
examples = []
f = open('adult.test', 'r')
for line in f:
line = line.strip('\n').strip('.').split(',')
example = get_input(line)
examples.append(example)
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'dnn' #your model_name which you set in docker container
request.input.example_list.examples.extend(examples)
response = stub.Classify(request, 20.0)
for index in range(len(examples)):
print(index)
max_class = max(response.result.classifications[index].classes,
key=lambda c: c.score)
re = response.result.classifications[index]
print(max_class.label,
max_class.score) # the prediction class and probability
def query(channel, examples):
# 发送请求
# request = predict_pb2.PredictRequest()
request = classification_pb2.ClassificationRequest()
'''
tensorflow_model_server --port=9000 --model_name=wide_deep --model_base_path=/tmp/pycharm_project_717/
rec_sys/recall_strategy/wide_deep-hh/wide_deep_model/model/wide_deep_64_40_2_1556356457
signature_def['serving_default']:
The given SavedModel SignatureDef contains the following input(s):
inputs['inputs'] tensor_info:
dtype: DT_STRING
shape: (-1)
name: input_example_tensor:0
The given SavedModel SignatureDef contains the following output(s):
outputs['classes'] tensor_info:
dtype: DT_STRING
shape: (-1, 2)
name: head/Tile:0
outputs['scores'] tensor_info:
dtype: DT_FLOAT
shape: (-1, 2)
name: head/predictions/probabilities:0
Method name is: tensorflow/serving/classify
'''
request.model_spec.name = "wide_deep" # the model name used above when starting TF serving
request.model_spec.signature_name = "serving_default" # get this from Saved Model CLI output
stub = prediction_service_pb2_grpc.PredictionServiceStub(channel)
request.input.example_list.examples.extend(examples)
response = stub.Classify(
request, 10.0) # 10 is the timeout in seconds, but its blazing fast
print(dir(response))
predictions = list(response.result.classifications)
'''
单个example输出记录：
classes {
label: "0"
score: 0.46758782863616943
}
classes {
label: "1"
score: 0.5324121713638306
}
# predictions[0].classes
[label: "0"
score: 0.7774916887283325
, label: "1"
score: 0.22250832617282867
]
'''
prob = [
round(predictions[i].classes[1].score, 4)
for i in range(len(predictions))
]
return prob
def testIssueRequests_RaiseRpcErrorIfRpcFailed(self):
# Prepare client and a side effect.
request = classification_pb2.ClassificationRequest()
client = TensorFlowServingClient('localhost:1234', 'a_model_name')
self.prediction_stub.Classify.side_effect = grpc.RpcError
# Call.
with self.assertRaises(grpc.RpcError):
client.IssueRequests([request])
def Request(text, ngrams):
text = text_utils.TokenizeText(text)
ngrams = None
if ngrams is not None:
ngrams_list = text_utils.ParseNgramsOpts(ngrams)
ngrams = text_utils.GenerateNgrams(text, ngrams_list)
example = inputs.BuildTextExample(text, ngrams=ngrams)
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'default'
request.model_spec.signature_name = 'proba'
request.input.example_list.examples.extend([example])
return request
def main():
parser = argparse.ArgumentParser(description='CTR model gRPC client')
parser.add_argument(
'tf_server',
type=str,
help='host:port for CTR Model TendorFlow Server')
parser.add_argument(
'chopstick_length',
type=float,
help='chopstick length to classify')
parser.add_argument(
'--model-name',
type=str,
default='tf_model',
dest='model_name',
help='model name to use')
parser.add_argument(
'--verbose',
'-v',
action='store_true',
default=False,
dest='verbose',
help='verbose output')
args = parser.parse_args()
host, port = args.tf_server.split(':')
channel = implementations.insecure_channel(host, int(port))
stub = prediction_service_pb2.beta_create_PredictionService_stub(channel)
# We use predefined ClassificationRequest protobuf here. This API is useful
# when you are working with tf.estimator package.
# tf.estimator.export.ServingInputReceiver expects to revieve TFExample
# serialized into string. All serialization and deserialization code is
# handled for us with gRPC and ClassificationRequest/ServingInputReceiver
# API.
#
# Consider using tf.make_tensor_proto function and
# tf.saved_model.builder.SavedModelBuilder if you are not using
# tf.estimator API
request = classification_pb2.ClassificationRequest()
request.model_spec.name = args.model_name
example = request.input.example_list.examples.add()
example.features.feature['Chopstick.Length'].float_list.value\
.append(args.chopstick_length)
result = stub.Classify(request, 10.0) # 10 secs timeout
print(result)
def parse_request(self, serialized_data):
request_fn_map = {
PREDICT: lambda: predict_pb2.PredictRequest(),
INFERENCE: lambda: inference_pb2.MultiInferenceRequest(),
CLASSIFY: lambda: classification_pb2.ClassificationRequest(),
REGRESSION: lambda: regression_pb2.RegressionRequest()
}
request = request_fn_map[self.prediction_type]()
request.ParseFromString(serialized_data)
return request
def main(_):
host, port = FLAGS.server.split(':')
channel = implementations.insecure_channel(host, int(port))
stub = prediction_service_pb2.beta_create_PredictionService_stub(channel)
request = classification_pb2.ClassificationRequest()
example = request.input.example_list.examples.add()
example.features.feature['petal_length'].float_list.value.extend([0.8])
example.features.feature['petal_width'].float_list.value.extend([0.8])
result = stub.Classify(request, 10.0) # 10 secs timeout
print(result)
def _create_classification_request(self, data):
if isinstance(data, classification_pb2.ClassificationRequest):
return data
request = classification_pb2.ClassificationRequest()
request.model_spec.name = self.model_name
request.model_spec.signature_name = self.signature_name
feature_dict_list = self._create_feature_dict_list(data)
examples = [_create_tf_example(feature_dict) for feature_dict in feature_dict_list]
request.input.example_list.examples.extend(examples)
return request
def _ExampleToClassificationRequest(
self, example: tf.train.Example
) -> classification_pb2.ClassificationRequest:
"""Convert single Example to ClassificationRequest.
Args:
example: `Example` instance to convert.
Returns:
A converted `ClassificationRequest` instance.
"""
request = classification_pb2.ClassificationRequest()
request.model_spec.name = self._model_name
request.model_spec.signature_name = self._signature_name
request.input.example_list.examples.append(example)
return request
def main(_):
host, port = FLAGS.server.split(':')
channel = implementations.insecure_channel(host, int(port))
stub = prediction_service_pb2.beta_create_PredictionService_stub(channel)
# Send request
# See prediction_service.proto for gRPC request/response details.
# Loop throught the images in folder
dir_path = path
# Read the directories available
list_files = os.listdir(path)
labels = []
images = []
f = open("answer.csv", "w+")
# Label each folder with a value starting
# from zero
for i in list_files:
print i + " loading ..."
print i.split(".")
f.write(i + "\n")
sample_image1 = iR2G.img2gs_res(test_image, 28, 28)
new_samples = np.array([sample_image1], dtype=np.float32)
image = new_samples.reshape(784, )
print(image.shape)
# Feature_dict
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'cat_and_dog'
request.model_spec.signature_name = 'serving_default'
example = request.input.example_list.examples.add()
example.features.feature['x'].float_list.value.extend(image.astype(float))
result = stub.Classify(request, 10.0) # 10 secs timeout
label0_value = result.result.classifications[0].classes[0].score
label1_value = result.result.classifications[0].classes[1].score
if (label0_value > label1_value):
print(0)
else:
print(1)
def call_servo(examples, serving_bundle):
"""Send an RPC request to the Servomatic prediction service.
Args:
examples: A list of examples that matches the model spec.
serving_bundle: A `ServingBundle` object that contains the information to
make the serving request.
Returns:
A ClassificationResponse or RegressionResponse proto.
"""
parsed_url = urlparse('http://' + serving_bundle.inference_address)
channel = implementations.insecure_channel(parsed_url.hostname,
parsed_url.port)
stub = prediction_service_pb2.beta_create_PredictionService_stub(channel)
if serving_bundle.use_predict:
request = predict_pb2.PredictRequest()
elif serving_bundle.model_type == 'classification':
request = classification_pb2.ClassificationRequest()
else:
request = regression_pb2.RegressionRequest()
request.model_spec.name = serving_bundle.model_name
if serving_bundle.model_version is not None:
request.model_spec.version.value = serving_bundle.model_version
if serving_bundle.signature is not None:
request.model_spec.signature_name = serving_bundle.signature
if serving_bundle.use_predict:
# tf.compat.v1 API used here to convert tf.example into proto. This
# utility file is bundled in the witwidget pip package which has a dep
# on TensorFlow.
request.inputs[serving_bundle.predict_input_tensor].CopyFrom(
tf.compat.v1.make_tensor_proto(
values=[ex.SerializeToString() for ex in examples],
dtype=types_pb2.DT_STRING))
else:
request.input.example_list.examples.extend(examples)
if serving_bundle.use_predict:
return common_utils.convert_predict_response(
stub.Predict(request, 30.0), serving_bundle) # 30 secs timeout
elif serving_bundle.model_type == 'classification':
return stub.Classify(request, 30.0) # 30 secs timeout
else:
return stub.Regress(request, 30.0) # 30 secs timeout
def prepare_classify_requests(instances, model_name, model_version):
request = classification_pb2.ClassificationRequest()
request.model_spec.name = model_name
if model_version is not None:
request.model_spec.version = model_version
instance_examples = []
for instance in instances:
feature_dict = {}
for key, value in instance.items():
if not isinstance(value, list):
value = [value]
feature_dict[key] = from_data_to_feature(np.array(value).ravel())
instance_examples.append(tf.train.Example(features=tf.train.Features(feature=feature_dict)))
request.input.CopyFrom(input_pb2.Input(example_list=input_pb2.ExampleList(examples=instance_examples)))
return request
def run():
token = 'Bearer eyJhbGciOiJSUzI1NiIsImtpZCI6ImxlZ2FjeS10b2tlbi1rZXkiLCJ0eXAiOiJKV1QifQ.eyJqdGkiOiI4Yzk3YWZjM2I4Mzk0N2E5OWVhMjMyNzE3Y2U3ZTFjMyIsInN1YiI6Im1scHRlc3RjbGllbnQiLCJhdXRob3JpdGllcyI6WyJtbHB0ZXN0Y2xpZW50Il0sInNjb3BlIjpbIm1scHRlc3RjbGllbnQiXSwiY2xpZW50X2lkIjoibWxwdGVzdGNsaWVudCIsImNpZCI6Im1scHRlc3RjbGllbnQiLCJhenAiOiJtbHB0ZXN0Y2xpZW50IiwiZ3JhbnRfdHlwZSI6ImNsaWVudF9jcmVkZW50aWFscyIsInJldl9zaWciOiJiNThjNWQ0ZiIsImlhdCI6MTQ4NjYzNDIzMSwiZXhwIjozNjMxMDgyMjMxLCJpc3MiOiJodHRwOi8vbG9jYWxob3N0OjgwODAvdWFhL29hdXRoL3Rva2VuIiwiemlkIjoidWFhIiwiYXVkIjpbIm1scHRlc3RjbGllbnQiXX0.nIl71Dxktizfb5B870Mlh_-62kN9_Wlda8WYbiz3iFaj22LzIUkQiRIAI57g3IwPXbJnJ1tlrf5_DIJpycRxzfxIZnW_GJW56sgY5L4mdPVHSIUHjeFh5v5tGwmOG6a1mYH_H0y8G-nHNolfSejcyvc4RYvcba4kS2nm-wDKKgfqDVaspM4Ktsa15eLHYn1P0LIUEsewTDm3qL_PgbJC3WKq_qgk02B5Or1n0doLkGBtccYlQEZ9lRixmkdov7_4Nl9UNTPgaYchC0AEaxd_RRCBK78FwC6tw3v1X3xJFXoYdJlMNOnTGdbQ4CVP5-Jd7gifPnUilPPPoJmITg0HZQ'
metadata = [('authorization', token)]
channel = grpc.insecure_channel("127.0.0.1:9000")
stub = tensorflow__serving_dot_apis_dot_prediction_service__pb2.PredictionServiceStub(
channel)
request = tensorflow__serving_dot_apis_dot_classification__pb2.ClassificationRequest(
)
request.model_spec.name = "func.pkl"
request.model_spec.signature_name = "classify_x_to_y"
example = request.input.example_list.examples.add()
example.features.feature["sepal length (cm)"].float_list.value.extend(
[5.1, 7.0, 6.3, 4.6])
example.features.feature["sepal width (cm)"].float_list.value.extend(
[3.5, 3.2, 3.3, 3.1])
example.features.feature["petal length (cm)"].float_list.value.extend(
[1.4, 4.7, 6.0, 1.5])
example.features.feature["petal width (cm)"].float_list.value.extend(
[0.2, 1.4, 2.5, 0.2])
response = stub.Classify(request)
print(response)
def run(host, port, input_str, model, signature_name):
channel = implementations.insecure_channel(host, port)
stub = prediction_service_pb2.beta_create_PredictionService_stub(channel)
# Pre-processing input
prediction_input = [json.dumps(eval(input_str))]
ink, classname = creat.parse_line(prediction_input[0])
# encapsulate as tf.Example object
classnames = ['doodle', 'expression', 'symbols']
features = {}
features["class_index"] = tf.train.Feature(int64_list=tf.train.Int64List(
value=[classnames.index("doodle")]))
features["ink"] = tf.train.Feature(float_list=tf.train.FloatList(
value=ink.flatten()))
features["shape"] = tf.train.Feature(int64_list=tf.train.Int64List(
value=ink.shape))
f = tf.train.Features(feature=features)
example = tf.train.Example(features=f)
final_req = [example]
start = time.time()
#generate request
request = classification_pb2.ClassificationRequest()
request.model_spec.name = model
request.model_spec.signature_name = signature_name
request.input.example_list.examples.extend(final_req)
result = stub.Classify(request, 10.0)
end = time.time()
time_diff = end - start
print(result)
print('time elapased: {}'.format(time_diff))
def wdl_sort_service(reco_set, temp, hbu):
"""
wide&deep进行排序预测
:return:
"""
# 1.读取用户特征中心特征
try:
user_feature = eval(
hbu.get_table_row('ctr_feature_user',
'{}'.format(temp.user_id).encode(),
'channel:{}'.format(temp.channel_id).encode()))
logger.info(
"{} INFO get user user_id:{} channel:{} profile data".format(
datetime.now().strftime('%Y-%m-%d %H:%M:%S'), temp.user_id,
temp.channel_id))
except Exception as e:
user_feature = []
if user_feature:
# 2.读取文章特征中心特征
examples = []
for article_id in reco_set:
try:
article_feature = eval(
hbu.get_table_row(
'ctr_feature_article',
'{}'.format(article_id).encode(),
'article:{}'.format(article_id).encode()))
except Exception as e:
article_feature = [0.0] * 111
# 3.serving服务端每一篇文章的example样本构造,要按照训练样本时的顺序和名字
channel_id = int(article_feature[0])
vector = np.mean(article_feature[11:])
user_weights = np.mean(user_feature)
article_weights = np.mean(article_feature[1:11])
# 组建example
example = tf.train.Example(features=tf.train.Features(
feature={
'channel_id':
tf.train.Feature(int64_list=tf.train.Int64List(
value=[channel_id])),
'vector':
tf.train.Feature(float_list=tf.train.FloatList(
value=[vector])),
'user_weights':
tf.train.Feature(float_list=tf.train.FloatList(
value=[user_weights])),
'article_weights':
tf.train.Feature(float_list=tf.train.FloatList(
value=[article_weights]))
}))
examples.append(example)
# 模型服务调用
with grpc.insecure_channel("0.0.0.0:8500") as channel:
stub = prediction_service_pb2_grpc.PredictionServiceStub(channel)
# 用准备好的example封装成request请求
# 并指定要调用的模型名称
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'wdl'
request.input.example_list.examples.extend(examples)
# 获取结果
response = stub.Classify(request, 10.0)
# 将结果与article_id组合，并按点击率大小排序，推荐出去
num = len(reco_set)
l = []
for i in range(num):
label_1 = max(response.result.classifications[i].classes,
key=lambda c: c.label)
l.append(label_1.score)
res = sorted(list(zip(reco_set, l)),
key=lambda c: c[1],
reverse=True)
article_list = [index[0] for index in res]
# 将排名前100个文章id返回给用户推荐
if len(article_list) > 100:
article_list = article_list[:100]
reco_set = list(map(int, article_list))
return reco_set
def wdl_sort_service(reco_set, temp, hbu):
#def wdl_sort_service():
"""
wide&deep进行排序预测
:param reco_set:
:param temp:
:param hbu:
:return:
"""
#hbu = HBaseUtils(pool)
# 排序
# 1、读取用户特征中心特征
try:
user_feature = eval(
hbu.get_table_row(
'ctr_feature_user',
#'{}'.format(1105093883106164736).encode(),
#'channel:{}'.format(18).encode()))
'{}'.format(temp.user_id).encode(),
'channel:{}'.format(temp.channel_id).encode()))
# logger.info("{} INFO get user user_id:{} channel:{} profile data".format(
# datetime.now().strftime('%Y-%m-%d %H:%M:%S'), temp.user_id, temp.channel_id))
print(user_feature)
except Exception as e:
user_feature = []
if user_feature:
# 2、读取文章特征中心特征
result = []
# examples
examples = []
#for article_id in [108245, 108246, 108242, 108240]:
for article_id in reco_set:
try:
article_feature = eval(
hbu.get_table_row(
'ctr_feature_article',
'{}'.format(article_id).encode(),
'article:{}'.format(article_id).encode()))
except Exception as e:
article_feature = [0.0] * 111
# article_feature结构： [channel, 10weights, 100vector]
# 构造每一个文章与用户的example结构,训练样本顺序
channel_id = int(article_feature[0])
vector = np.mean(article_feature[11:])
user_weights = np.mean(user_feature)
article_weights = np.mean(article_feature[1:11])
# 封装到example(一次一个样本)
example = tf.train.Example(features=tf.train.Features(
feature={
"channel_id":
tf.train.Feature(int64_list=tf.train.Int64List(
value=[channel_id])),
"vector":
tf.train.Feature(float_list=tf.train.FloatList(
value=[vector])),
'user_weights':
tf.train.Feature(float_list=tf.train.FloatList(
value=[user_weights])),
'article_weights':
tf.train.Feature(float_list=tf.train.FloatList(
value=[article_weights])),
}))
examples.append(example)
# 所有的样本，放入一个列表中
# 调用tensorflow serving的模型服务
with grpc.insecure_channel("127.0.0.1:8500") as channel:
stub = prediction_service_pb2_grpc.PredictionServiceStub(channel)
# 构造请求
# 模型名称，将要预测的example样本列表
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'wdl'
request.input.example_list.examples.extend(examples)
# 发送请求
response = stub.Classify(request, 10.0)
print(response)
def transform_view():
from flask import request
csv = request.files['file']
csv.save("Network_Test_Traffic.csv")
chunksize = 23
chunk_list = []
missing_values = ["n/a", "na", "--", "Infinity", "infinity", "Nan", "NaN"]
for chunk in pd.read_csv("Network_Test_Traffic.csv",
chunksize=chunksize,
na_values=missing_values):
chunk_list.append(chunk)
break
dataFrme = pd.concat(chunk_list)
lstcols = []
for i in dataFrme.columns:
i = str(i).replace(' ', '').replace('/', '')
lstcols.append(i)
dataFrme.columns = lstcols
dataFrme = dataFrme[lstPred_col]
df_predict_original = dataFrme.copy()
scaler = MinMaxScaler(feature_range=(0, 1))
selected_X = pd.DataFrame(scaler.fit_transform(dataFrme),
columns=dataFrme.columns,
index=dataFrme.index)
df_predict1 = selected_X.copy()
max_rows = 2800
dataframes = []
dataframes_original = []
while len(df_predict1) > max_rows:
top_original = df_predict_original[:max_rows]
top = df_predict1[:max_rows]
dataframes.append(top)
dataframes_original.append(top_original)
df_predict1 = df_predict1[max_rows:]
df_predict_original = df_predict_original[max_rows:]
else:
dataframes.append(df_predict1)
dataframes_original.append(df_predict_original)
final_df = pd.DataFrame(columns=lstPred_col)
for i, j in zip(dataframes, dataframes_original):
j.index = pd.RangeIndex(len(i.index))
j.index = range(len(i.index))
examples = []
for index, row in i.iterrows():
example = tf.train.Example()
for col, value in row.iteritems():
example.features.feature[col].float_list.value.append(value)
examples.append(example)
channel = grpc.insecure_channel(server)
stub = prediction_service_pb2_grpc.PredictionServiceStub(channel)
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'Model_Network'
request.model_spec.signature_name = 'serving_default'
request.input.example_list.examples.extend(examples)
response = stub.Classify(request, 10.0)
testdata = df_predict1[lstPred_col]
outputs = testdata.copy()
for index, row in outputs.iterrows():
max_class = max(response.result.classifications[index].classes,
key=lambda c: c.score)
outputs.loc[index, 'Location'] = get_key(int(max_class.label))
outputs.loc[index, 'Probability'] = max_class.score
print(outputs)
final_df = pd.DataFrame(columns=[
'BwdIATMean', 'BwdIATTot', 'BwdPktLenMax', 'BwdPktLenMean',
'FlowDuration', 'FlowIATMean', 'FlowIATStd', 'FwdPSHFlags',
'FwdSegSizeMin', 'InitBwdWinByts', 'Location', 'Probability'
])
final_df = final_df.append(outputs, ignore_index=True)
os.remove("Network_Test_Traffic.csv")
return render_template('view.html',
tables=[final_df.to_html()],
titles=['na'])
def wdl_sort_service():
"""
wide&deep进行排序预测
:param reco_set:
:param temp:
:param hbu:
:return:
"""
hbu = HBaseUtils(pool)
# 排序
# 1、读取用户特征中心特征 1115629498121846784
try:
user_feature = eval(hbu.get_table_row('ctr_feature_user',
'{}'.format(1113244157343694848).encode(),
'channel:{}'.format(18).encode()))
# logger.info("{} INFO get user user_id:{} channel:{} profile data".format(
# datetime.now().strftime('%Y-%m-%d %H:%M:%S'), temp.user_id, temp.channel_id))
except Exception as e:
user_feature = []
if user_feature:
# 2、读取文章特征中心特征
result = []
# examples
examples = []
# for article_id in [17749, 17748, 44371, 44368]:
for article_id in [22324, 22325, 22326, 22327]:
try:
article_feature = eval(hbu.get_table_row('ctr_feature_article',
'{}'.format(article_id).encode(),
'article:{}'.format(article_id).encode()))
except Exception as e:
article_feature = [0.0] * 111
channel_id = int(article_feature[0])
# 求出后面若干向量的平均值
vector = np.mean(article_feature[11:])
# 第三个用户权重特征10维
user_feature = np.mean(user_feature)
# 第四个文章权重特征10维
article_feature = np.mean(article_feature[1:11])
# 组建example
example = tf.train.Example(features=tf.train.Features(feature={
"channel_id": tf.train.Feature(int64_list=tf.train.Int64List(value=[channel_id])),
"vector": tf.train.Feature(float_list=tf.train.FloatList(value=[vector])),
'user_weigths': tf.train.Feature(float_list=tf.train.FloatList(value=[user_feature])),
'article_weights': tf.train.Feature(float_list=tf.train.FloatList(value=[article_feature])),
}))
examples.append(example)
with grpc.insecure_channel('127.0.0.1:8500') as channel:
# 建立连接通道
stub = prediction_service_pb2_grpc.PredictionServiceStub(channel)
# 获取测试数据集，并转换成 Example 实例
# 构造 RPC 请求，指定模型名称。
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'wdl'
request.input.example_list.examples.extend(examples)
# 发送请求并获取结果
response = stub.Classify(request, 10.0)
print(response)
return None
def wdl_sort_service():
"""
wide&deep进行排序预测
:param reco_set:
:param temp:
:param hbu:
"""
hbu = HBaseUtils(pool)
# 排序
# 1、读取用户特征中心特征
try:
user_feature = eval(hbu.get_table_row('ctr_feature_user',
'{}'.format(1115629498121846784).encode(),
'channel:{}'.format(18).encode()))
# logger.info("{} INFO get user user_id:{} channel:{} profile data".format(
# datetime.now().strftime('%Y-%m-%d %H:%M:%S'), temp.user_id, temp.channel_id))
except Exception as e:
user_feature = []
if user_feature:
# 2、读取文章特征中心特征
result = []
# examples
examples = []
for article_id in [17749, 17748, 44371, 44368]:
try:
article_feature = eval(hbu.get_table_row('ctr_feature_article',
'{}'.format(article_id).encode(),
'article:{}'.format(article_id).encode()))
except Exception as e:
article_feature = [0.0] * 111
# article_feature结构： [channel, 10weights, 100vector]
# 构造每一个文章与用户的example结构：和 wide_and_deep.py 中训练样本顺序，格式也必须相同（求了平均）
channel_id = int(article_feature[0])
vector = np.mean(article_feature[11:]) # 和训练时相同：求平均（简化工作）
user_weights = np.mean(user_feature) # 和训练时相同：求平均（简化工作）
article_weights = np.mean(article_feature[1:11]) # 和训练时相同：求平均（简化工作）
# 封装到example(一次一个样本)
example = tf.train.Example(features=tf.train.Features(feature={
"channel_id": tf.train.Feature(int64_list=tf.train.Int64List(value=[channel_id])),
"vector": tf.train.Feature(float_list=tf.train.FloatList(value=[vector])),
'user_weights': tf.train.Feature(float_list=tf.train.FloatList(value=[user_weights])),
'article_weights': tf.train.Feature(float_list=tf.train.FloatList(value=[article_weights])),
}))
examples.append(example)
# 所有的样本，放入一个列表中
# 调用tensorflow serving的模型服务
with grpc.insecure_channel("127.0.0.1:8500") as channel:
stub = prediction_service_pb2_grpc.PredictionServiceStub(channel)
# 构造请求
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'wdl' # 模型名称
request.input.example_list.examples.extend(examples) # 将要预测的example样本列表
# 发送请求：获取结果
response = stub.Classify(request, 10.0)
print(response)
# 是要返回 response的预测结果的，但是现在没有运行，不知道response的数据结构，所以只能暂时返回None
return None
# -*- coding: utf-8 -*-
from tensorflow_serving.apis import classification_pb2
from tensorflow_serving.apis import regression_pb2
from tensorflow_serving.apis import predict_pb2
from tensorflow_serving.apis import prediction_service_pb2
from grpc.beta import implementations
'''
pip install tensorflow-serving-api
pip install grpcio
'''
channel = implementations.insecure_channel("0.0.0.0", int(8500))
stub = prediction_service_pb2.beta_create_PredictionService_stub(channel)
request = classification_pb2.ClassificationRequest()
example = request.input.example_list.examples.add()
#example.features.feature['SepalLength'].float_list.value.extend([float(100)])
result = stub.Classify(request, 10.0) # 10 secs timeout
print(result)
def transform_view():
request_file = request.files['data_file']
request_file.save("iBeacon_RSSI_Unlabeled.csv")
TF_MODEL_SERVER_HOST = os.getenv("TF_MODEL_SERVER_HOST", "127.0.0.1")
TF_MODEL_SERVER_PORT = int(os.getenv("TF_MODEL_SERVER_PORT", 9000))
BLE_RSSI_UL = pd.read_csv('iBeacon_RSSI_Unlabeled.csv')
COLUMNS = list(BLE_RSSI_UL.columns)
FEATURES = COLUMNS[2:]
LABEL = [COLUMNS[0]]
df_predict = pd.read_csv('iBeacon_RSSI_Unlabeled.csv')
df_predict = df_predict.drop(['date','location'],axis = 1)
df_predict_original = df_predict.copy()
df_predict[FEATURES] = (df_predict[FEATURES] - df_predict[FEATURES].mean())/df_predict[FEATURES].std()
max_rows = 2800
dataframes = []
dataframes_original = []
while len(df_predict) > max_rows:
top_original = df_predict_original[:max_rows]
top = df_predict[:max_rows]
dataframes.append(top)
dataframes_original.append(top_original)
df_predict = df_predict[max_rows:]
df_predict_original = df_predict_original[max_rows:]
else:
dataframes.append(df_predict)
dataframes_original.append(df_predict_original)
final_df = pd.DataFrame(columns=['b3001','b3002','b3003','b3004','b3005','b3006','b3007','b3008','b3009','b3010','b3011','b3012','b3013','Location','Probability'])
for i,j in zip(dataframes,dataframes_original):
j.index = pd.RangeIndex(len(i.index))
j.index = range(len(i.index))
examples = []
for index, row in i.iterrows():
example = tf.train.Example()
for col, value in row.iteritems():
example.features.feature[col].float_list.value.append(value)
examples.append(example)
channel = implementations.insecure_channel(TF_MODEL_SERVER_HOST, TF_MODEL_SERVER_PORT)
stub = prediction_service_pb2.beta_create_PredictionService_stub(channel)
request1 = classification_pb2.ClassificationRequest()
request1.model_spec.name = "blerssi"
request1.model_spec.signature_name = "serving_default"
request1.input.example_list.examples.extend(examples)
response = stub.Classify(request1, 10.0)
outputs = j.copy()
for index, row in outputs.iterrows():
max_class = max(response.result.classifications[index].classes, key=lambda c: c.score)
outputs.loc[index, 'Location'] = get_key(int(max_class.label))
outputs.loc[index, 'Probability'] = max_class.score
final_df = final_df.append(outputs,ignore_index=True)
return render_template('view.html',tables=[final_df.to_html()],titles = ['na'])
def transform_view():
from flask import request
csv = request.files['file']
csv.save("iBeacon_RSSI_Unlabeled.csv")
BLE_RSSI_UL = pd.read_csv("iBeacon_RSSI_Unlabeled.csv",
encoding='utf8') # Unlabeled dataset
COLUMNS = list(BLE_RSSI_UL.columns)
FEATURES = COLUMNS[2:]
LABEL = [COLUMNS[0]]
# Data Preprocesssing
df_predict = BLE_RSSI_UL # Unlabeled dataset
df_predict = df_predict.drop(['date', 'location'], axis=1)
df_predict_original = df_predict.copy()
# df_predict[FEATURES] = (df_predict[FEATURES] - df_predict[FEATURES].mean()) / df_predict[FEATURES].std()
df_predict[FEATURES] = (df_predict[FEATURES]) / (-200)
max_rows = 2800
dataframes = []
dataframes_original = []
while len(df_predict) > max_rows:
top_original = df_predict_original[:max_rows]
top = df_predict[:max_rows]
dataframes.append(top)
dataframes_original.append(top_original)
df_predict = df_predict[max_rows:]
df_predict_original = df_predict_original[max_rows:]
else:
dataframes.append(df_predict)
dataframes_original.append(df_predict_original)
# server = str(TF_MODEL_SERVER_HOST)+":"+str(TF_MODEL_SERVER_PORT)
final_df = pd.DataFrame(columns=[
'b3001', 'b3002', 'b3003', 'b3004', 'b3005', 'b3006', 'b3007', 'b3008',
'b3009', 'b3010', 'b3011', 'b3012', 'b3013', 'Location', 'Probability'
])
for i, j in zip(dataframes, dataframes_original):
j.index = pd.RangeIndex(len(i.index))
j.index = range(len(i.index))
examples = []
for index, row in i.iterrows():
example = tf.train.Example()
for col, value in row.iteritems():
example.features.feature[col].float_list.value.append(value)
examples.append(example)
# channel = grpc.insecure_channel(server)
stub = prediction_service_pb2_grpc.PredictionServiceStub(channel)
request = classification_pb2.ClassificationRequest()
request.model_spec.name = 'Model_Blerssi'
request.model_spec.signature_name = 'serving_default'
request.input.example_list.examples.extend(examples)
response = stub.Classify(request, 10.0)
outputs = j.copy()
for index, row in outputs.iterrows():
max_class = max(response.result.classifications[index].classes,
key=lambda c: c.score)
outputs.loc[index, 'Location'] = get_key(int(max_class.label))
outputs.loc[index, 'Probability'] = max_class.score
print(outputs)
final_df = final_df.append(outputs, ignore_index=True)
os.remove("iBeacon_RSSI_Unlabeled.csv")
return render_template('view.html',
tables=[final_df.to_html()],
titles=['na'])